Show us your half baked, not really ready for prime time projects.
Also, if you need any help with a project, a startup, or an idea, just post it here.
https://news.ycombinator.com/item?id=1822253 – Oct 23, 2010, 307 comments
https://news.ycombinator.com/item?id=25700135 – Jan 9, 2021, 821 comments
FXcursion is an STM32-based guitar processor prototype. It has 3 stereo inputs and outputs, builtin effects chain, looper and recording. It lacks proper user interface (we've found a designer last week, so it's in the works now!), and the hardware still needs a lot of changes, but I've been impressed with the progress so far.
https://github.com/eimrine/resume/blob/main/eimrine-s%20rese...
One has to enter some abstract of what they want to write, and some text they have so far, and the tool generates possible continuations.
It requires quite a bit of editing, maybe 3-4 attempts for each completion if one wants to generate decent text. Feedback was mostly positive from a dozen or so professional writers who tried it out for a day.
I feel that it generates more naturally looking text than ChatGPT with the typical prompts.
See my profile if you want to try it out. You can use the demo without registration.
A website to create digital invitations. Similar purpose as WhatsApp groups for invites, but taking away the noise and adding way more personalized design to the invite.
The UI still has lots of issues and I need to i18n the landing page, but it works and a couple of people use it every day already. I‘m spending 2€ per day on google ads to see how my changes to product alter user behavior.
It’s a side project and I‘m not planning to squeeze money out of it. Just want the world to celebrate more and experience building a product from zero to „actively used by many people“.
Because if it does then I have https://peopleandblogs.com
I thought it’d be cool to have a small textual language where you could describe “this payment goes out every quarter” and “this one goes out monthly from date X to Y” and so on. As a freelancer it can be difficult to know how much cash you have available given taxes go out at weird times, and it’s useful to simulate “let’s delete this row where the customer pays” to stress test your finances and so on.
I also wanted it to be “privacy-first” because I figured I wouldn’t trust my financial data to a random person on the internet, so neither would I expect anyone else to.
So I created this site and put it online but I’m a developer, not a marketer. I did a “Show HN”, it got some nice comments but not much traction.
Not sure if I should spend effort marketing it. It’s just a free site so I’m only doing this for fun. Or maybe the product is wrong and needs changes e.g. more features.
Difficult to know what to do. Advice appreciated!
Esther is my personal project to develop a diary app that talks back. Basically. Along the way I learned how to htmx, llama.cpp via JNA and practical experience with LLMs. I suffer from type-1 bipolar disorder with psychoses and normal life just ain't gonna happen. So unemployed, single, #nolife ... I decided to build a chat bot for myself. It's not finished the RAG stuff still needs implementing, but oh well. It's written in Clojure on GraalVM (because I wanted to try polyglot) other than that it's nothing /too/ fancy ... but it feels great having an LLM + UI you can just build the way you like it
View, search, and analyze arbitrary source code (best support of Swift right now) in 3D and AR space. You open your phone or tablet, yeet hundreds of files into 3D space, and can start highlighting, moving, and tracing execution by literally walking around your code. The desktop app has similar features, and the standard 3D viewer is just as fun.
I would love help - from anyone of any kind - to build this out towards greater usefulness. It’s a lot of fun, it’s super cool to look at, and it’s the thing I’ve wanted to use since I was a small child.
“Let me touch the words!!” P.s - the dictionary and word wall samples are fun too
This is a NSFW site that uses randomization. I'm trying to work towards some of the magic that StumbleUpon used to have.
This project I made a while ago and never took it further: https://github.com/weinberg/concurrencyRunner. Concurreny Runner allows you to debug multiple running processes simultaneously, with breakpoints set to allow specific interleaved execution paths to be explored. This allows you to trigger concurrency scenarios which are typically very difficult to analyze because they rely on random timing. That repo has examples which demonstrate typical database concurrency issues of "read skew", "write skew" and "read modify write". I envisioned it as something you could run in CI to actively test for these things like we do other integration testing.
This other one is in the sad graveyard of promising projects which I never could devote enough time to: https://forums.tigsource.com/index.php?topic=69545.0. It's an adventure game where I recorded hundreds of panoramic images and stitched them all together into a seamless walkable player experience. One day...
It's been almost 2.5 years of work, it supports uptime monitoring for websites, APIs, JavaScript apps, cron job monitoring, has built-in status pages, still extremely far from "baked".
Slightly fancier paste bin with markdown support that I want to turn into a group editing service that takes inspiration from pull request style workflows.
Just as a warning, not even half of the stuff on the landing page are actually present in Coral but they absolutely will be future. The next thing on the plan is to re-write the UI in React Native so I can take Coral with me on the go.
https://coral.nectarine.sh (the download page is inactive, but you can grab a release from the GitHub page if you're so inclined)
I'm building a service that helps founders (just like you and me) craft their investor updates in literally seconds. We're currently in the testing phase of the MVP, so I could really use some help.
I'd love to give anyone of you a demo of the tool and see if it ACTUALLY alleviates your pain related to any part of creating and sending investor emails, and if I'm not chasing windmills, haha.
Let me know if you're up for a chat sometime next week. 15min max, promise! https://calendly.com/n10v thanks
FHIR data viewer (a healthcare data format if you're new to FHIR). I've been working on it for some time now, and just released the Mac build a couple of weeks ago.
Had a 'soft launch' on LinkedIn a week ago that was very well received by my followers.
Still a lot of work to do before a commercial release - huge features missing. But last week's launch was positive - a real boost.
https://www.linkedin.com/feed/update/urn:li:activity:7114844...
My Show HN was a flop. :)
https://logictrader.xyz - Online crypto trader, but still needs a lot of work. I'm a bit short on time now with the AGI project.
Alternative Wifi app to connect to Fujifilm cameras. I mainly wanted to just try and make a MVP to see if it's possible, and now it's grown into something usable.
I built this for personal use, but maybe someone else also finds it useful.
It's a Firefox add-on that arranges YouTube video tabs based on the runtime of the video. I often hoard many YouTube tabs and at some point I want to either watch the shortest ones or play something lengthy so I can do household chores while listening to the video. This makes finding the correct video from tens of different tabs so much easier. There is one known major bug: if a video is playing, the sorting doesn't work.
There is an older version published for Microsoft Edge, but Edge Add-ons started rejecting the updates based on unclear reasons (something along the lines of "no value for user" ???). When reaching for support through email, I only get responses from people who don't understand English and just copy-paste the exact same unclear rejection report and close the ticket.
Chrome Web Store has a publishing fee and outright rejects me paying it, probably because I live in Switzerland and my credit card is from a Finnish bank.
In the tsunami of music today, Remuse is a way to sift through the sounds and discover quality contributions. Musicians, bands, and fans are invited to create a premium account which gives you co-editor status on Remuse.co allowing you to add tracks to our feed.
The Premuse page shows tracks before they are on the main page, [premium accounts](https://remuse.co/create-account) can cast votes and write track descriptions for tracks on the main page.
We syndicate our publications to Twitter and more, we have a YouTube podcast to go over fresh tracks that we feature, and we send out [The Remuseletter](https://letter.remuse.co) with high quality fresh music content, track descriptions written from a musician's perspective, and beautiful album art, directly to your inbox [3].
Discover fresh music before your friends, Remuse is a place for bands and artists to get heard.
We are in Phase A of the launch sequence and by getting a premium account now you can really help us achieve our goals and support our mission to ensure that artists and musicians do not starve. We're partnering with a local nonprofit to hold an incubator gala in November (next month) to help fund musicians on their trajectories and empower the music scene in our area.
[2] https://letter.remuse.co (The Remuseletter -- Free Fresh Music to your Inbox)
[3] https://remuse.co/create-account (Musician and Listener Premium Signup, get co-editor status)
Obsidian Local LLM Assistant. It's actually a fork of someone else's project who was working with the Open AI API.
https://www.ytemail.com/ - e-mail notifications for YouTube uploads. I was in the 0.01% that used and liked e-mails, and built this after Google killed the e-mails. If you want to use, please get in touch, I'm happy to onboard free users if the amount of notifications you'll get isn't big.
https://github.com/kassner/whattocook - Recipe chooser, but it works the other way around you'd expect. In a household of depressed people, it's common that we can't get to agree on the meal, so this project decides it for us. You can exclude ingredients in case you don't have them at home, but that's the only way you'll get a different recipe.
https://www.kassner.com.br/projects/money/ - Personal finances project that only I use. Never had the courage to open it to the world, and there isn't anything innovative in it to be worth the hassle either. Only pluses are single-binary web project and DBs are password-encrypted by default.
Game projects okay? More a toy at the moment, it's just the factorio tech tree but I'm working to have my own recipes and story and features.
Started as a testbed for learning typescript + React, grew into something to work on when I'm feeling creative.
Code is more likely to be updated at https://gitlab.com/omgbear/factorymustgrow, but GitHub pages is easy.
The half baked bits at the moment is full scale automated catalyst discovery, and a bunch of the "glue" code that pulls the different simulations together seamlessly (molecular dynamics, dimer methods, nudged elastic band, retrosynthesis, automated bulk analysis) -- they all exist in various states of development (mostly done) but getting them all to interact totally seamlessly is tricky.
Also this is quite computationally intensive hobby, I am often limited by the amount of computing power I have - even though I run a small cluster in my house (384 cpu cores, 4 GPUs, 24TB NAS)
I am super excited though, just in the last week I have started to make novel, truely unique catalyst discoveries (that are matching lab validations), so after a year of experimenting "it's working.gif"
It's fun, and works great for our purposes. Nothing can compete with it, in my book. Unfortunately, we haven't found a customer worth pleasing and, since we are well-pleased ourselves, we haven't found a reason to approach full-baked.
Edit: for demonstration purposes, "Who is the greatest band of all time?" https://druthers.app/#/de7ef4ea-35e6-4d0f-9b10-2945129eba2e
AI provides key decision considerations and recommends the best option while your team benefits from real-time polling, pro/con analysis, and decision archiving.
This is something like Teleport, but open source and easy to host yourself, with clear extension points and a heavy focus on ergonomics.
I have lots of ideas around this, but it’s hard to pour all the time into a side project for me right now :(
The pitch: Want to run stuff on Raspberry Pi, but dread having to remember how to set it all up again when the sdcard fails? Try this homebrew-like interface for common stuff.
Backed by docker + docker-compose + a template system. Not really a package manager but has similar qualities.
really need to somehow get airlines to give this a code review and try this in real life somehow.
Whipped up a quick way to play videos in sync across different computers so people can watch movies together, remotely.
Basically, I think that LLMs will enable a whole new set of app UXes, and I'm trying to build a platform for those UXes. In a sense, a "shell" for LLM apps. It's a command-bar style UI with an SDK that makes it really easy to build functionality on top of LLMs / vector-dbs etc and to interact with other software/files (e.g. VSCode, Chrome, etc.). Currently very limited docs but if you're interested in building LLM workflows/tools, I'd love to collab!
It's getting there. Can almost run a ticker plant w/ real money on Alpaca.
https://github.com/halftheopposite/graph-dungeon-generator: what was supposed to be circular graph dungeon generator, ended up being a tree dungeon generator.
https://webcursors.click - I should have spent more time thinking about solving a problem that doesn't exist and clearly you can feel that there's something missing.
[0] https://github.com/RiftValleySoftware/RVS_IPAddress
[1] https://github.com/RiftValleySoftware/RVS_ParseXMLDuration
[2] https://github.com/RiftValleySoftware/RVS_ONVIF
This project is unfinished (I just walked away from it, as it wasn't really giving me what I wanted):
[3] https://github.com/RiftValleySoftware/RVS_GTDriver
This one is "half-baked," I believe. I never really took it particularly far:
Im reimplementing RandomX (Monero's PoW) algorithm trying to optimize its original and improved (the one used by miners software) versions. All of that while learning cpp20/23.
It helps you get access to someone's Google ads, Analytics, Tag manager, Search console, GMB and a few others in one click.
- source code: https://github.com/agi-merge/waggle-dance
- buggy demo available @ https://waggledance.ai
- really neat (IMO) adaptation of Plan-Validate-Solve
- similar to AutoGPT and its ilk but with different architecture decisions - seeking co-founders/contributors
My last evening project was a virtual business card webapp. Cards can be designed with web technology and you can drag to rotate to see the back-side :)
It's much further along than what is available online right now.
All data and tables update in real-time and support complex views including filtered by category and/or based on variables. It's designed to be fast and scalable.
It will soon be possible to use as a flexible CMS for websites but the next phase is to add access control to allow building forms to collect data from users.
It's a generative art platform/marketplace. You write an algo that generates art according to some number of parameters, upload it as a Docker image, and let other people tinker and generate things they like the aesthetics of. Basically remote code execution as a service, but I spent a lot of time on the sandboxing side of things and feel pretty good about it.
The infra is tuned for "saving money" and not "scale", so there's a chance it falls over under even modest load. If folks are interested I can tweak it a bit to make it more scale-y.
It's also my HN name, because all the other (one word) ones I could think of were taken.
Work in progress on an IDE for 3D graphics written in Common Lisp.
Cheesy trailers:
https://Earsmith.com is a "completed" (but unpublicized and probably only used by me) project I'm proud of — it's for training your pitch sense.
I find that most of what I do, on any given day, is finding interesting stuff and sending it to people to see what they have to say about it. Hackernews is pretty effective at doing this, so it feels like i’m more or less building hackernews but with power-user features for collecting and sharing stuff.
I imagine in the future being able to group content and then synthesize it with AI to get a draft of something you have been wanting to write.
https://maedia.page/84b4dcec/4961380-rapper-tom-macdonald-ca...
https://maedia.page/84b4dcec/8023848-confessions-of-the-prof...
I'd say I'm about 80% done with the project, but is taking longer than expected for certain specifications I wanted to include.
https://github.com/ballsteve/xrust
It's two of us trying to improve Rusts XML offering, by implementing an XSLT processor and all that entails. I'm working on the parser at the moment, I have DTD entity expansion working but still need to figure out UTF-16 support and validation of DTDs. I'm probably going to focus on XML ID support next.
Any help, feedback, suggestions would be appreciated!
It is a companion site for a second project to convert old format (images, videos and audios) to digital. Converted media end up in the first site This is half baked at https://www.audiovideotransfers.co.uk/
This is a python-syntax re-imagining of Ansible / Cookiecutter: declarative IT automation and project templating. It's fairly early, but the heart of it is there: I'm able to create files/directories and render templates, running it displays status, I have most of the "magic" implemented that makes it less like Python and more like Ansible. The CLI entry-point is minimal, I need to basically forklift over the uPlaybook1 argument handling code and playbook search/selection.
Would love feedback from someone who is familiar with Python and Ansible about the direction of the YAML -> Python for playbook syntax, but I worry that it's too early to even give that.
Basically it's uPlaybook (v1) but with the Ansible-inspired YAML playbooks reimagined as Python: https://github.com/linsomniac/uplaybook
I recently released this recipe search engine focused on finding recipes with your target macros.
I’m switching gears to focus on a real startup, but this was a fun little project that’ll never see real completion. I got the basics of a meal planner working (this was a cool problem to solve) but there was a massive amount of data labeling work required to take it to the next level.
https://github.com/arilotter/vaportrade https://vaportrade.net/
A realtime data tool for operational teams. Hoping to make sense of BPMN processes.
A free tool for recruiters to find talented software developers online.
I am really proud of this one. Will continue working on it sometime. idk.
Local-first personal task management with estimation.
https://uuidv7.app is an online tool which generates UUIDv7 compliant ids. I also provide a simple free api at [1] which returns something like this: `{"id":"018b25e4-78a3-75e2-93cf-4004aef6c2ae"}`. Please be kind, there's currently no rate limiting.
No ads, works perfectly without javascript, a tool for fellow programmers.
I'll soon add a feature to extract the timestamp from the id as a debugging too.
It’s still an early beta but feedback is welcomed!
https://chrome.google.com/webstore/detail/skysavvy-enhance-g...
It's a Zelda-like game engine with some 1000ish games made in it over 20 years. Used to be called Zelda Classic, but we renamed it this year.
I especially enjoyed implementing the navigation sidebar for the release notes (you need a wide enough screen to see it): https://zquestclassic.com/releases/2.55-alpha-119/
https://github.com/shi-yan/codestage
a chatgpt powered vocabulary book (for generating example sentences and quiz)
I quit my job at the bank after a bad experience. With no source of income I started to develop this game. After 3 days in, I thought to myself "who am I kidding? I won't be able to sell this and make money" so I went job hunting and found the job I work today (3 years in).
Although it looks like it only have one level, it doesn't. I wrote a level generator for it.
I did spend a lot of time working on things that don't matter, for instance screen orientation. My goal was to have one game that i could sell in any platform, mobile or desktop.
Maybe I'll finish it someday.
https://github.com/meltyness/cef-tex
The build system is a VS file with god-knows-what hardcoded, doesn't properly reference CEF or Godot, and is therefore locked to Windows platform only (though Godot can theoretically run everywhere), and is probably full of a bunch of other footguns that I haven't even thought of. Got 5 stars on GitHub just from tag discovery, and the wishful thinking of some Godot-ers that I'll ever have time to un-suck it. It's also got R-rated comments, a video demonstration, and a short written log of the weird hurdles I had to surmount.
I didn’t like the existing solutions, so I decided to create my own. In active development now. I’d say 60% is done.
My attempt to get more out of all my ebook highlights using on-device AI.
https://github.com/dmotz/trystero
My attempt at making it as easy as possible to add p2p to a web app (no server required). A "jQuery for WebRTC" of sorts.
Basically a simpler version of ros. Easily, and performantly (is that a word?) connect cameras and other sensors, via xtensor/eigen/numpy to whatever algorithms you have, and control actuators/robots.
Does meal planning and grocery lists, but want it to suggest dinners and tell me when the food I bought is going to go off.
https://github.com/demyinn00/BootMe
BootMe is a workflow manager app. There are 6 buttons that can be configured. 4 of the 6 buttons can be configured to open tabs and links, launch apps, and start a spotify playlist. This app is almost ready, but I'm looking for feedback. Specifically, bugs and potential enhancements.
Please do not roast my UI, I'm sure there are improvements that can be made. Offer suggestions and constructive criticism. For the record, I'm a new grad -- go easy.
Thanks!
These are my hardware/robotic/vision/navigation related projects. They are pretty noobish but fun to work on.
Unfortunately lately I haven't had as much time as I'd like to work on this stuff.
If anyone is interested in looking as far as providing architecture/code review type thoughts, I'd be open to them.
[1] autofocus camera with speech intent intended later for tracking [2] visual depth probe slam from crude panorama/"color segmentation" [3] single point slam insect-style quad
[1] https://github.com/jdc-cunningham/ml-hat-cam [2] https://github.com/jdc-cunningham/pi-zero-2-robot-navigation... [3] https://github.com/jdc-cunningham/twerk-lidar-robot/tree/dev
An open source offline-first browser DB that uses any S3 API as the cloud storage. Backblaze, minio, S3, R2 in the test harness. Randomised causal consistency testing behind toxiproxy. 30kb but still hardening the offline-first stuff and only GET and SET KV operation implmented so far (but multiplayer), next step will be conditional SET as a step toward transaction support. All over vanilla S3!
I tried to extend the idea to cargo build (see http://alsuren.github.io/2022/07/30/why-cargo-quickbuild.htm...) but I never got it to work with proc macro packages so I gave up. If anyone wants to have a go at getting it working, I'm happy to mentor.
Still need to add a bunch of challenges, find a way of monetise it, and remove the waiting list (need some more guard lines first). I really enjoyed working on it with NextJS, nodeJS with fastify, and vclusters to create sandboxed playgrounds.
I started coding it early 2023, but took a break from it in the last 3,4 months because... well summer.
https://git.sr.ht/~jamesponddotco/privytar
Still need to write tests and deploy an official instance, but the code itself should be working already. I'll consider it ready when those tasks are done and I can have a senior Go developer review the code.
I was inspired by projects like Nitter and Bibliogram. No real reason for it to exist aside from that and "because I wanted to." I guess I also don't trust Automattic all that much either.
There's also acopw, a small cryptographically secure password generator module for Go:
https://git.sr.ht/~jamesponddotco/acopw-go
I'll consider it ready once the underlying cryptographic implementations have been independently audited.
It lets you manage EFI boot entries in a TOML file.
It's actually in pretty good shape, but I still want to work out a reproducible and offline build system, for the sake of packaging in Debian, Arch, etc.
An interpreter for AMPscript (DSL for personalizing emails) that I used when I was at ExactTarget/Salesforce. I wanted to learn Rust and needed a project-based approach.
It doesn't really have a real-world use case since it's not integrated with any email platform (e.g., Marketing Cloud). I had loads of fun working on it though! Oh, and it's still missing a garbage collector.
Turn your iPad into a jukebox.
I’ve thrown 20+ parties with it and it’s always a hit. Putting it out in the universe for fellow jukebox nerds to try out too!
It’s functional programming for creative thought processes.
I’m excited to launch this very soon, and hope you’ll check it out. Thanks!
It's a more efficient accounting ledger targeted at in-house finance teams of post Series A type startups.
It's a tool to make shell scripting read a little more like English.
I'm developing an platform that allows the user to create generative art directly in the UI using pre-programmed patterns (only three at the moment).
In short, this Ghidra extension allows one to reconstruct relocation tables through analysis and then export parts of programs as working object files, effectively reversing the work of a linker. Applications include binary patching, converting between object file formats, software ports without source code, decompilation projects...
I've been tinkering with it for the past 16 months or so and it's the third, hopefully industrial-grade prototype. Right now it can delink 32-bit MIPS and i386 programs from the 1990s or so to ELF object files, as long as it contains basic relocation types.
It's half-baked because while it works, it doesn't support modern instruction sets, advanced relocation types for TLS/PLT/GOT or exporting to other object file formats besides ELF, so it's not that useful on modern artifacts (which is what I assume most reverse-engineers would care about). It's not really ready for prime time because I'm not done writing blog posts that walk through real-world application and case studies ; there's very little literature out there on this esoteric topic and it can be very confusing. Like "let's take this PlayStation PS-EXE file that was built with a COFF toolchain back in the 90s and make MIPS ELF object files out of it that work with modern Linux toolchains" kind of confusing.
I started this project because I wanted to decompile a PlayStation video game and quickly realized that I'd never get anywhere without a means to divide and conquer it into more manageable pieces. Ironically the decompilation project itself hasn't advanced much, but I'm having fun so far working on this.
I move it slowly in my personal computer but not much in public. Maybe adding another person will help me on that!
App that lets you crowdsource people's opinions on your photos. Unlike other websites out there, it doesn't rate you 1-10 but instead ranks your photos against each other so you don't have to feel bad about yourself :). Very useful for dating apps and other social media, esp. if you don't have easy access to friends' opinions.
I've had lots of friends and users tell me it's super useful, but I'm in a slump on building it right now.
I made a little thing to count and track how many people are surfing on surfline's webcams to learn a bit about ml/ai, computer vision and deploying these things. I think its neat, and the data is fun in a nerdy way. Not sure what to do with it now. Its been cool, I had to get my own dataset together for it to work.
Check it out at and click around dp52hy5y2zsyt.cloudfront.net
I haven't committed enough to it to buy a domain.
Nutrition balancer
>The Halfbakery is a communal database of original, fictitious inventions, edited by its users. It was created by people who like to speculate, both as a form of satire and as a form of creative expression. (To learn more about what ideas are appropriate, see the help section.) The halfbakery software is implemented as one big C CGI program, edited with vi, compiled with gcc, invoked by an Apache http server.
After recognizing the pretty steep learning curve many users face with Anki, including myself, I started working on munkle.it as a more intuitive alternative. The goal is to create the app that streamlines the flashcard creation process, prioritizing user-friendly design and simplicity, making it easier for learners to dive straight into effective study sessions without the initial overwhelm.
I want loosely coupled, functional, primarily no dependencies, practical, typescript first (if something can't be expressed in safe ts, it's too dynamic to belong there), easy to use and hackable/extendable packages. If they don't do i/o they need to work from browser ie. from observable, they're all dual esm/cjs.
Have plans for incorporating pretty much all AI image gen. capabilities, but only has image gen, loras, controlnet, background removal and upscale atm. There's like something new in AI image gen coming up everyday and I still haven't gotten lora training, checkpoint training, image colorizing, AnimateDiff and so much more done yet.
Dreadful name, and definitely half baked.
Stick in some text in English and have it translated into a bunch of different languages and then generate audio of that same text in WAV format.
Node graphs, OSINT data mining, and plugins. Connect unstructured and public data for transformative insights. My long-term goal is to turn this project into a viable alternative of Maltego/Palantir type software.
Currently my roadmap looks something like:
- Refactor and clean up the code base all over to make it easier to onboard potential contributors
- Improve the UI/UX
- Moving/sandboxing the plugin system/crawlers to it own compose service
- Adding a settings option to the plugins system and a way to directly query a graph with gremlin
- Adding another graph display mode to the UI (using https://www.sigmajs.org/)
- Creating a registry/marketplace for plugins where anyone can upload, share, and download plugins
- Adding accounts and RBAC
- And many more things to come but for now the above tasks are my focus :)
Encrypted messages without history - for when you want to send an api key or password or the like and don't want it sitting in the recipients WhatsApp/Signal history, for example, or going over non-encrypted channels like email or Slack. And you don't want to set up a complicated encryption messaging thing. There are similar services already: e.g. onetimesecret, burnernote - but I wanted something that did encryption client side. I also wanted to set up channels, to send multiple messages between 2 devices without doing "trust on first use" for each new message.
The encryption takes place in the browser - if you're a developer: you can even use browser devtools debugging/network sections to confirm the private key never leaves your device.
I plan to work on the ux, and also further strengthen the encryption (e.g. does the signal protocol work for this...? will implement something that should provide some form of forward secrecy in any case), and make a Flutter-based client so it's super convenient to use on all platforms (maybe also enable things like nfc-based key change...).
Would be very interested to hear from anyone who would like to help! :)
This is a drawing toy that takes some inspiration from kidpix. It makes amusing (to me anyhow) noises, has some goofy brushes and stamps.
You can't save your pictures or print them out, and I haven't gotten around to figuring out some happy medium of using canvas and doing a flood fill.
haha, and trying it on my laptop, something is wrong with an offset somewhere or other. ah well, half-baked it is
Costream allows you to enjoy livestreamed content across multiple platforms simultaneously.
Professional e-sports tend to be limited to the official broadcast. This has created a subculture in the streaming world where people mute the main broadcast but continue to watch it, and then open a second stream of a professional player and listen to their commentary.
Costream aims to enable that with customizable UI elements.
I have a ton of features in the pipeline, but I've decided to take a break and wait to see how many users are actually interested in it before dedicating more time.
I've been working on building a content intelligence dataset on podcast data. Using new LLMs I'm able to transcribe, speaker detect, and topic extract anything in the podcast space (or audio in general) and produce a detailed structured dataset from what people are talking about.
The website is jank as hell atm, but I think the resulting transcripts are pretty interesting - https://siev.io/episode/b6139bf3-f680-42c0-bc9d-f195fd71f293
You use it to break the deterministic timeline your life is stuck on. However, it only works if you always abide by outcome no matter what.
It's a small go web server that queues jobs (commands in a whitelist and a list of args) and runs them in go routines with some max parallelism, and a browser extension to submit commands with the URL as the last argument. I use it to run scraping and archiving scripts on things mostly.
My current side project, a Computer Assisted Translation tool targeted at comics, manga, web toons, etc
https://shred.willsmithte.com/
Right now, it's just a web landing page. I have a mobile app prototype I'm testing with some people, especially focusing on the Vancouver area.
A sort of Cargo/NPM/poetry for C++.
I don't have a lot of confidence in Conan and I see people struggle with CMake. This project combines a lot of ideas from both of those and hopefully makes things easier by basically combining the two into one tool.
It has some overlap with Meson but I didn't like particularly like Meson either.
Build recipes are written in Starlark - a subset of Python with a focus on immutability also used by Bazel and Buck2.
It can currently generate build files for Ninja and MSVC. There's no reason it couldn't also generate for XCode and Make I just haven't done that yet.
It's still at a proof-of-concept stage as I test out ideas.
iOS open source app that is an AWS dictionary and a logo repository’s for people drawing AWS diagrams.
A virtual pet for Arduino: Feed, cuddle, clean and educate your own unique cat
2 mini games, 7 food types, friends visits and presents, one button action, no sound, no death.
I made a YouTube summarizer, Summary Cat.
————————-
Technical Details
Goal: Generate brief and coherent summaries from YouTube video transcripts
Motivation of building this project: to save myself time watching videos that are too long.
Stack • Frontend: HTML/CSS, plus a tiny bit of Javascript for handling the button • Backend: Python/Flask • Flask served as the backend web framework, handling routes and integrating various services. • Front+Backend has about 500 lines in total • Hosting: AWS Free-tier Elastic Beanstalk • Database: AWS Free-tier RDS
APIs: • For grabbing YouTube's transcripts: I used youtube-transcript-api (https://pypi.org/project/youtube-transcript-api/) • For summarizing the transcripts: I used OpenAI's GPT-3.5-turbo-16k: https://platform.openai.com/docs/guides/gpt. • I used GPT-3.5 because GPT-4 is quite a lot more expensive (roughly 10X).
My Prompt (Super Simple!) • "please summarize the following text into a few paragraphs:" + the full transcript.
Thoughts about GPT-4 vs GPT-3.5-Turbo-16k or Summary Cat • GPT-4 was 20% better for "summary quality" • GPT-4 feels 50% faster • However, GPT-4 is about 10X as expensive as GPT-3.5
https://github.com/MeoMix/symbiants
I'm making a digital ant farm. SimAnt + Tamagotchi that runs in real-time. Written in Rust/WASM/Bevy
The goal is to give people cute ants that do cute ant things 24/7, ask people to look after their ants and feed them once-per-day, and then deny the ability to provide ant care unless users engage in a breathwork/meditation exercise.
I would love help. I'm very much so in over my head. I have a billion things to add and the number one demotivator is not having people to tinker with, bounce ideas off of, and get hyped alongside. https://discord.gg/pg5Tu68cdW come say hi!
A weird and cumbersome web based synthesizer.
In the examples below, double click on a node to trigger a signal in the circuit and produce sound.
Here is a "song" that a friend made with it: https://alex.miller.garden/sonic-circuit/?2.9.C4.m.1.15,6.9....
And here is "Happy Birthday": https://alex.miller.garden/sonic-circuit/?3.1.G3.m.1.24,4.1....
It shows you semantically relevant code context around the lines that match your search. This lets you see the loops, functions, methods, classes, etc that contain all the matching lines. You can get a sense of what's inside a matched class or function definition. It shows relevant code from every layer of the abstract syntax tree, above and below the matches.
https://github.com/paul-gauthier/grep-ast
Here's an example, which would also be colorized in most terminals:
$ grep-ast encoding
│#!/usr/bin/env python
│
│import argparse
⋮...
│def main():
│ # Parse command line arguments
│ parser = argparse.ArgumentParser()
│ parser.add_argument("pattern", nargs="?", help="the pattern to search for")
│ parser.add_argument("filenames", nargs="*", help="the files to display", default=".")
parser.add_argument("--encoding", default="utf8", help="file encoding")
│ parser.add_argument("--languages", action="store_true", help="show supported languages")
⋮...
│ for fname in enumerate_files(args.filenames):
│ process_filename(fname, args)
│
⋮...
│def process_filename(filename, args):
│ try:
with open(filename, "r", encoding=args.encoding) as file:
│ code = file.read()
│ except UnicodeDecodeError:
⋮...
│ print()
│
⋮...
│if __name__ == "__main__":
│ res = main()
│ sys.exit(res)it’s a cli utility tool to manage/replace .env files.
I was tired of having 3 different .env files in every project folder. They fall out of sync when you’re developing stuff and you only update your .env.local file, only to realise, later on, that you did not update the values in your .env.stage and all the others too.
This cli leverages an sqlite database to abstract away these files. Instead of having multiple .env you can create as many environments as you want (local, stage, prod etc.) and each one can contain multiple key-value elements. It has some nice commands already but I plan to make some more useful stuff with it.
It works, yeah, but I'm just not getting around on actually learning CAD to design a nice case for it. Also I started another project that takes this idea further, supports multiple switches, expression pedals, acts as HIDs, etc. but I keep getting distracted by starting even more "projects" of course.
It works great but promoting it was a big fail.
Typocide is a CLI based typing tutor. It solves a few problems which other typing tutors failed to provide for me. In particular it uses real world text instead of fake text (such as keybr). The corpus is derived from all the quotes available on Typeracer. It also does not follow the same rules as 10fastfingers or monkeytype's default mode of displaying words in random order. To make an analogy with music, when I practice a passage in a piece of music, I don't play the measures in a random order. Rather, I practice challenging _phrases_ of music (groups of related measures).
Next, Typocide draws inspiration from leveltype (https://github.com/christoofar/leveltype) which bans the use of the backspace key. If you make a typo, Typocide will identified the typo'd word as the focus. It will then build a test which includes the surrounding words for the current test (previous and next word), and also the surrounding words from previous tests where you have typo'd the word.
The idea is that when you make a typo, the word was a part of some context, and rather than simply practicing the word, you want to practice the word in context. A lot of typos, particularly in speedtyping, happen at word boundaries. Furthermore, by having it operate on context, it also reinforces common idioms. For example, the word "example" is very likely to be preceded by the word "for" and very unlikely to be preceded by a word like "kaleidoscope". So it makes sense to practice "For example," as a unit.
Furthermore, it has some rudimentary support for identifying words which you have typed more slowly than usual. Typocide will also generate practice tests from these slow words.
Sometimes the tests are not perfect, so I have added a simple escape hatch: if you press TAB on a new test, it will skip the test, and if you press TAB after you have started typing a test, it will repeat the test from the beginning.
This project is half baked for two reasons: 1. The interface is very rudimentary. 2. The code is somewhat disorganized and needs to be cleaned up.
- Keyboard layout generator (ocaml) - https://github.com/rdavison/stronglytyped
This is a keyboard layout generator with support for multiple layers and punctuation optimization. This project is half baked because I can't figure out a set of weights to make it generate great layouts (it generates decent layouts, but not any layouts which are :chefskiss:). Also, at some point I would like to move the analysis away from dealing with floating point numbers and move it towards integers.
The design is quite interesting because it uses Jane Street's Incremental library to heavily cache intermediate results. As the generator is running, it tries random swaps and then re-evaluates the score. Oftentimes a swap will affect one part of the keyboard, but not the other... for instance if you swap two keys on the left hand, then it will not affect anything on the right hand. In that case, this tool will only recompute the parts that changed instead of recomputing the entire layout.
I still haven't done any benchmarking to see if this _actually_ improves performance, or whether it just makes me feel better. :)
Current E2E verifiable e-voting systems are complex to deploy and often require a trusted setup phase if fancy zero-knowledge proofs are used. All require coordination of threshold decryption ceremony. Moreover, maintaining bulletin board immutability is considered an easy problem; just replicate it. This necessitates that the e-voting systems are deployed with multiple independent parties that don’t collude, significantly adding to the deployment costs. Deployment can be delegated to a third party, but the tendency to optimise costs would sacrifice “independence” for the parties, which ensures the voter’s anonymity.
Some while ago, I got attached to the idea of anonymising voters’ pseudonyms instead of the votes with which they cast their votes pseudonymously. The anonymisation of the pseudonyms can be done verifiably on untrusted third parties transactionally - meaning when the resulting anonymisation is computed, returned and verified, there are no lingering trust assumptions. Another essential ingredient is a History enabled bulletin board, which provides efficient inclusion and consistency proofs. This enables voter’s client applications to ask bulletin board consistency proofs, ensuring their and others’ vote inclusion in the final tally. The result is a fully centralised e-voting system, PeaceFounder, which a single person can deploy without concerns about integrity or privacy.
More on it can be found on peacefounder.org
Ampwall. Musician-owned, Public Benefit Co alternative to Bandcamp. (Eventually.) https://ampwall.com. Dogfooding it with my own band’s merch sales at https://woeunholy.ampwall.com. Digital audio will launch soon (2-3 weeks?) and we plan on starting our tests with outside bands/labels by end of year.
I’ve been building this throughout the past year but the stuff that’s publicly available is pretty basic. I was going to keep it quiet but the news about Epic selling Bandcamp two weeks ago prompted me to announce early, so might as well share it here too and really put the cart before the horse!
I have a lot of opinions about things I’ve always wanted Bandcamp to offer. While it’s going to start very similar to BC, we’re going to experiment with some very different things ASAP. The feedback I gathered throughout the year showed me that there’s a hunger for this and the response to my announcement was bigger than I expected, so I’m feeling good about it. It’s been fun to build and it’s going to get better.
It’s all Next.js and TypeScript. Hosted on Node.js servers because I don’t think serverless is for me. I’m using Porter.run for production, a YC alum whose office is in my neighborhood in Queens!
A spreadsheet-based approach to food I've been using with my partner for years, turned into an app. Track only calories & protein, look at weeks instead of days, eat healthy/don't eat CRAP (calorie-rich and processed foods).
https://apps.apple.com/us/app/tarot-raven/id1539747669?platf...
https://github.com/ferg1e/comment-castles
https://www.commentcastles.org
For moderation, a user whitelist sits in front of the typical content blacklist, which introduces some interesting properties.
Webpage screenshots as a service. I was scratching a personal itch with this one but maybe some day I will make it something others can use at scale.
Backend powered by Google Cloud Run, Pupeteer, and headless Chrome. Website is Next.js.
StoryTeller is an exploratory web application that creates short audio stories for pre-school kids.
It used speech-to-text, llms, text-to-speech, embeddings and image generation.
StoryTeller is built with the following libraries:
ModelFusion Fastify Next.js shadcn/ui Zod
The following AI APIs are used:
OpenAI Eleven Labs Lmnt Stability
Convert your GitHub Gists into a full featured blog and RSS feed.
Example: https://gist.aviperl.me/avi-perl https://gist.aviperl.me/avi-perl/rss
By default it works as a simple redesign of an existing set of gists. But gist hosted configuration files and compiling gist data to a static site can turn it into something much more rich, seo friendly, etc.
Semantic data warehousing and analytics tool written in python. It has experimental/half-baked NLP features to query your warehouse by interacting with the semantic layer with AI, instead of the normal approach of having an LLM write SQL and needing to know your entire schema.
"Kayak for sneakers"
I wrote crawlers to pull product data from about 100 stores, post sales and filter on size availability.
Some important features are still missing and site could use some polish
An AI real estate listing generator. It's pretty functional for basic use cases not but needs a lot more work to handle edge cases.
Some of the features I am including in this version: - import music from the Music app (downloaded locally), from Files app, via a web server - a music visualizer, with three presets - waveform views - mark and store segments of a song - bookmarks - an equalizer and a bunch of smaller features, like pinning music to the Home view, remembering playback for an album or playlist.
TestFlight: https://testflight.apple.com/join/7WsEOF6u
It’s playable, but in order to consider it complete I’d need to add instructions, a start button, a game over screen, and a replay button. And I’d want to add a color blind option, sounds, and a way to share your result (score & a screenshot of your final state).
An internet radio aggregator & player featuring music radio stations organised by genre rather than location.
There are tonnes of great music focused and specialist stations out there. TuneIn or RadioGarden don’t really make it easy to find the best stations for a given genre though.
I wanted to build it as a web app that’s usable as a PWA on mobile but have had lots of issues streaming audio reliably.
Also had issues with audio cutting out on web.
I’m currently rebuilding as the web audio API has improved things. Not enough hours in the day as usual!
The website is for tracking betting lines and finding arbitrage and +EV bets across sports betting sites. I've been hooked on the idea that there are inefficiencies in sports betting markets that can be taken advantage of, especially in betting on live events and so this started as me just collecting minute by minute odds data from different sites and going from there.
I'm currently collecting all the data myself so my next step is creating a public API to allow other users to access the same sportsbook data that I'm currently using (https://sportsbookapi.com).
https://github.com/obaqueiro/dominoparty
The idea is to make a minimalistic tabletop online multiplayer game engine that only draws the "pieces" and controls the player-hidden pieces but doesnt enforce game logic.
I tried to simulate the environment I had when playing dominoes with my friends in real life.
The idea of the engine is that you can change the "pieces" (cards, letter squares, domino tiles) and thus can play lots of games online with friends.
I built it at the start of covid to play with the family. But it has A LOT of rough edges, bugs in state keeping and the code is awful
I'd love to have time to improve it. Particularly to keep state using CRDTs via Yjs.
I wrote a stoichiometry library in rust to practice my numerical and linear algebra programming. I wanted a way to share it, other than as a command line binary. (This had the added bonus of letting me practice writing a web server in rust)
I have wanted to improve the equation balancer by implementing the linear system solver explicitly "in the domain of the natural numbers" (currently it's in "reals" -- i.e. float64) , which should be intuitive given rusts type system, but I haven't found a straightforward way to use any of the existing libraries as such.
I'm the ArchiveBox.io creator, and I'm experimenting with different ideas to launch a paid version of ArchiveBox for less technical users. ArchiveBox will always remain free and open source, DigestBox would just be a fancy frontend on top of it that handles cloud hosting and provides a nicer UI to make internet archiving more approachable to the average user.
The revenue would go towards supporting open source ArchiveBox development, and would allow me to work on it full-time.
I'd love to get people's feedback on this concept. Let me know if you like the idea of DigestBox and if you'd use it!
I'm also curious if the $0.01/URL pricing sounds reasonable to people. Assuming each page produces 50mb or less of output, that works out to $120/yr to archive 1000 URLs per month.
- https://github.com/huytd/speech: Practice english speaking for non-native
- https://github.com/huytd/ascii-d: a cross platform ASCII diagram editor
- https://github.com/huytd/js-playground: a JS code playground with a little visualization
- ...
My wife is going through a cyber security program and needed to quickly get familiar with binary, decimal, hexadecimal conversion. So I made a little practice tool in React. It's been a few months since I did React so I had to take a couple hours to learn how you bootstrap React projects these days, then another couple hours whipping this up. More I want to do.
wzornik-etykiet.com
Of particular note - a janky filtering system for each label!
Also - no english, yet!
Other quirks: - Individual labels are not linkable, if you want to share one, the link won’t work! - No proper mobile support.
Honestly, this project was a learning experience in React for me, and the code is a mess. No back end, just JSONs and sheets behind it all.
I launched it today, it's a workout mini app within my collection of mini apps, it's still a WIP given that for now there's only a nice HIIT Timer and a workout history, all the other sections about exercises, tools and programs will come later.
In practice it's not your usual gym app with a knowledge base already included if stuff to do, here the scope is gonna be the ease of building all of it by yourself and owning/sharing it as you wish.
I wrote it to scratch my own itch so I can type faster on my Pinephone. I had no idea about writing Rust code, language models or Wayland protocols and ended up writing a master thesis about it. The language model used n-grams and was generated from Wikipedia. The idea was that everybody is able to easily generate a model for their own language. The keyboard worked fine, but the language model was way too big.
I got side tracked writing an input simulation library that works on all platforms. Once I am done with that, I'll get back to the keyboard and make it usable again.
A golf foursome scheduler app for your close friends and acquaintances. There’s weather api integration, notifications, and GolfNow deep linking. We’re expanding to golf influencers and public groups soon.
The core game mechanic is an endless word-finding game, where it is possible to save up letters to spell some very high scoring words. There is also a 5-minute mode for people who need an ending. Oh, and a 2-player versus mode that virtually no one wants to play, according to anonymous stats.
I'm still ironing out bugs and adding some features, but it's coming along.
An epub reader. It's still in development. But if what you need is only reading, then it just works.
I've been using Shipscoop for years now to the delight of many clients, but we've yet to finish its "go live" flow. That said if people want to try it out, it works quite well for turning git logs into weekly status reports
Search doctors that actually did research papers on the thing you want. EG mohs surgery.
https://www.opendoctor.io/research/?research_papers=mohs&zip...
I keep an eye on a number of project changelogs and needed to be notified when they update. Since they’re in various places, and don’t all have releases so I can’t use GitHub’s release notifications, I built a simple tool to consume from anywhere with an url and send me a diff. Slowly adding features on top of that.
I’d love to know if this is valuable to anyone else on the planet but me. Or what would make it so.
A free, open source, local-first, spaced repetition system that works offline, has p2p syncing, plugins, and first class support for collaboration. It's GitHub/Reddit for flashcards.
I basically took Anki and turned it into a webapp >_>
A Facebook-style ActivityPub server. Has wall posts, groups, and events.
Not yet "ready" — 1.0 would need much better moderation tools, photo albums (with tagging), discussion boards (in groups), and a client API. Direct messages are a work in progress right now.
I myself use it daily for my participation in the fediverse.
It's a web app to help you improve your real-life pool (billiards) game. You can practice shots and learn to visualize the angles. Eventually, I want to go way beyond basic shotmaking and teach patterns, but it's already useful in it's current state at least to beginners and intermediate players.
I've posted about it before but I've made a lot more progress since then, including making it mobile-friendly.
It's Wordle but multiplayer and each player only gets one guess. Right now it's just for my friend group, but I may try to open it up so you can make your own squad with your friends if I have the time.
a collaborative online web space for creative coding
the main feature is it saves the full html of the page on every change
this means you can make full web apps with only front-end code
My half arsed Chat-GPT web client. Rust (web-sys) and wasm
Good enough for me, I use it every day. But not even half an arse visually.
No I have it to the point I can use it I am moving on to real time music software
I wrote it because I was frustrated with OpenAI's Code Interpreter and I thought working in a real notebook would be better.
What's half-baked? Using the GPT4 API gets expensive if the conversation gets long enough. Also, the install isn't as easy as it could be, since you have to install the 'llm' command-line tool first, and set your API key.
It supports two languages, Python and TypeScript (via Deno), but Deno's notebook support is itself pretty half-backed.
https://marketplace.visualstudio.com/items?itemName=skybrian...
lovezone.io
The problem with commercial services is the short-term profit thinking caused by economic incentives. It is in the interest of the dating sites to hold on to paying customers as long possible. Each pairing will cause them to lose two customers.
Therefore, it is against the economic interest of these companies to really help people to get into relationships and find each other.
That is the reason why such sites show you only some of the profiles, but it seems that you do not necessarily see the profiles that would be a really good match for you. Somewhat cynically it seems that they show you profiles are used as a bait to keep you coming back to the site.
On the other hand, showing people endless number of profiles skews the attitude of the people. Why settle for getting to know this one person behind this one profile, when a better person might be just around the corner. This causes dating exhaustion, which is a real phenomenon. Under such stress, people start overemphasizing superficial things, such as the looks.
Lovezone is based on the observation that in the physical realm, mating often happens in smaller circles: within a group of friends or within hobby groups. Lovezone makes possible for such small circles to have their own dating site.
So Lovezone is less a dating site, and more a platform for making dating sites. The idea is that anybody can start a group, which is kind of a miniature dating site, in Lovezone in just few minutes, for free.
We host GitHub Actions runners for you, with faster CPUs and larger storage at lower prices than the official ones. Currently running a waitlist and onboarding customers in batch to validate the idea.
Something of a passion project that's being used by a few researchers now. We're hoping to get a paper out at some point. Currently overhauling the fitting code/config handling.
turdness.com
The site grew from my personal frustrations with Finland Post Corporation. In four months, three packages were returned and three others were almost returned, most from different parties. Apparently this was due to a systematic error. I did everything in my power to get them to fix it, but to no avail. They did not want to even acknowledge the issues exist. Everything was my fault.
I found some companions in this misfortune in the Facebook group Suomen Posti - Paska laitos (engl. Finland Post - Shitty Facility). After reading experiences in this group, and thinking about my own experience, I started thinking that maybe it is possible to classify organizations based on how irredeemable their "turdness" is:
Level 1: Core process fails
Level 2: Even backup process fails
Level 3: Failed to monitor success
Level 4: Customer service fails to help in case of failure
Level 5: Escalation via customer service fails to help
Level 6: Executive escalation fails to help
Level 7: Filing an external complaint to overseeing organization fails to help
Idea is that people can vent their frustrations about an organization, and we can collectively collect systematic data about the turdness of that organization. This could then potentially help the organization to overcome their turdness.
Couldn't find an easy setup uptime monitoring service with some advanced features, so i built one. The features i missed in existing tools:
- monitoring databases (sql, mongodb) with simple queries, checking rabbitmq queues, checking OS memory and disk space.
- monitoring private resources not exposed on the internet and/or on private networks.
No landing page ready yet. You can check out the demo here: https://www.loom.com/share/8814e6613a3c483487bd61beadef842b
What we offer:
- Fully white-labeled invoicing
- White-labeled Invoice PDFs
- White-labeled payment portals & domains
- Customized notifications & copy
- Customized ACH statement descriptors
- Use our same-day ACH & card payments or BYOPayments provider
Please check out the video and let me know if you have any questions or thoughts!
We're always looking for feedback (and early users!)
I'm also sort of releasing backwards in that I decided to charge up front to limit the number of users and slowly gather feedback while mitigating my risk, but once I work out all the kinks there will definitely be a free tier that everyone starts out on.
https://chromewebstore.google.com/detail/smudgeai/mhagofdace...
As a long time lifestyle design/self-dev enthusiast, i've been trying a lot of routines, workouts, etc... along the years and i keep forgetting good stuff.
So i built this tool to keep track of my routines, create (complex) routine schedules, and share/discover other people's routines
QR-powered home organization app! My family and friends use it constantly, but I never polished or promoted it further than that video :) If you're organizing your garage/shed/projects bench and want to give it a try, I'm happy to mail you a big stack of stickers!
It's not finished yet..
On Product Hunt: https://www.producthunt.com/products/unycorn
I quit my job more than a year ago and wanted to participate in this crazy (but fun) rush to find creative applications of AI. Tried many things, first attempts failed miserably, but it's a fun journey and I'm getting useful experience.
Right now, I'm building an app that allows you to apply AI to create interesting visual effects for images and videos.
Currently, there are two main features:
- You can swap faces in videos to create personalized memes.
- Generate artistic self-portraits.
You probably have seen such features many times already and maybe even tired of it, but for me it was just a starting point. I plan to continue adding new features as new interesting models and technologies become available. Right now there is only a web version and I'm working on rolling out ios and android versions (Almost there!).
Stack: AWS, docker, runpod.io, python, react native.
Happy to collaborate with other makers on this and other ideas!
It's built using Flutter and primarily relies on standard UI widgets.
Currently, it is available for free without any ads, and I'm working on adding more content that will be unlocked through in-app purchases.
All the art is generated using Midjourney. If it weren't for AI-generated art, I wouldn't have even started the project.
Try it:
Android: https://play.google.com/store/apps/details?id=com.cercaapp.g...
I hated having to open an excel spreadsheet after dinner with friends. Snap a picture of a receipt, send them the URL, and let them claim what they ordered.
OCR and ML is finally good enough to reliably itemize a receipt, even with variable templates and layouts.
It's not the fun stuff so I've been putting it off forever :/ Also kind of on a break from programming at the moment so not sure when I'll return to this.
I had a contract recently where I got very familiar with render.com, and I figured I'd do some blogging about some of the things I learned, with some tips and tricks to learn how to write technical content.
It's been a lot of fun to help people on the internet with something I've learned a lot about!
I'm working on an adapter to plug a USB controller into an Atari 2600. The above video shows my first working proof of concept. Currently only supports HID devices with a fixed button mapping, and acts like a joystick controller.
It isn't complete, but I'm using it for my site. It needs some help.
1. https://nodesteaders.com/ (Community of Future-Forward Homesteaders)
2. https://typezebra.com/ (Simple google fonts tester)
3. https://atlasgeopolitica.com/ (poor man's CIA factbook)
Cartobucket Auth is an OIDC/OAuth 2.1 system that is meant to be easy to extend and embed into your existing infrastructure. To that end, if you are using Java, you can use the built in auth-data package, you can also use the generated SDK's for the API. Cartobucket Auth is also built to be extended or parts replaced using the Quarkus DI system easily, should you need the system to do something else that is specific to your application.
You can view a system diagram here: https://s.icepanel.io/sCSXIaaJOAsrpe/qAMx
Find the best episodes for any podcast and recommend your favorite episodes. Reconciling different podcast episodes was a slog and maybe the hardest technical thing I've ever worked on. I never officially gave up on it, I've just been putting off coming back to it.
Minimal blogging app with a judgement-free writing mode, focused on getting you to write more. Actively working on this one!
Started as ChatGPT over email -> then messaging ( SMS/discord/whats app etc. ) -> chat embed / configurable AI bot REST API. Been all over, not sure what to do next to be honest!
Nap is a command-line API testing tool that uses YAML files. I wrote it because I couldn’t find anything equal to it. I use it at work to help our QA folks build test suites for our back-end services.
Multiplayer HTML 3D worlds.
Served its initial purpose, but the domain will expire in 60 days. Not sure if worth the renewal.
My first synthesizer. Web components using Web audio and Web midi apis, but no library dependencies. It's built to be extensible, and able to create new synth instruments by providing a bunch of source elements (like audio or video tags use, with custom properties), but those features aren't deployed yet. So just a regular old synthesizer that works in any browser (no midi in Firefox, though), for the time being!
The repos can be fed as list on CLI, via file, or via a "source" type plugin (allows github search or custom API calls), the PR creation works via a "Forge" plugin.
Resulting PR list can be fetched to review status:
> Of 113 given PRs, 73 merged, 30 closed not merged, 5 ready to merge (no conflict), 5 with conflicts.
I've created hundreds of PRs with it via github app integration, see the plugins repo[2].
[1]: https://jiby.tech/mass-driver/usage.html [2]: https://github.com/OverkillGuy/mass-driver-plugins.html
A charting / note-taking text editor.
Tab to get autocomplete suggestions. Highlight and click the summarize button to summarize text.
Bookmarking / read later with automatic archiving and “for you” recommendations from the small web based on saved articles.
It's an in-house shipping protection platform for Shopify stores. 260 million packages are lost/stolen per year, and it's really not on the brand in a lot of cases. Huge hit on margin and when given the choice, a lot of customers prefer the piece of mind.
My goal is to give brands the ability to run this in house instead of relying on a third party that could give a worst possible outcome for brand and customer.
Want to make it have an AI claims agent and build a bunch of workflows. The widget took forever to pull it off though so we ship what we have!
Note to self, embeddable widgets are hard no matter how simple they look.
https://garden.simonsarris.com/
It all works (double click to place plants, undo/redo, copying, etc etc), but it needs more items. And saving and loading does not work yet. Actually, since this is HN, you could write
go.Diagram.fromDiv(myDiagramDiv).model.toJson()
Some examples of what I'm thinking so far to give an idea of what loading might look like: https://twitter.com/simonsarris/status/1705056223333151053
I have considered making a more public version of this that anyone can use for their own garden, but that would take more non-trivial work.
A static sites generator which uses JavaScript instead of a template language. It evaluates the javascript in your pages ahead of time.
Basically web analytics but using websockets so we can track time spent on each page. Then there is a daily email report which is helpful for monitoring launching new products.
I never got any paying customers so I switched to other stuff. Marketing is hard. Someday I'll go back to it. I use it though and it works, but the interactive reporting is half baked.
I'm also working on https://sidewaysdata.com - Motorsports registration and timing. New design will be up soon (done by actual designer, not me lol)
It’s nothing special, but I like it a bit better than the official MTA app in NYC :)
started as a mint / personal capital / spreadsheet replacement. Thought it would be neat to have a grafana-like way to build visualizations on aggregated accounts. I started with investments b/c it's easy to input/import without linking to bank accounts. I have Plaid development access and think a flexible playground will be more interesting for expenses.
It's an app to help 2+ people choose what to watch together on the most popular streaming platforms by presenting semi-random titles, asking each person to vote (swipe left/right), and settling on the "best" based on most votes/popularity.
It was born out of endless scrolling through Netflix (and the like) with significant others and friends, and not settling on something to watch. Instead, this shortens the process by allowing everyone to give their input, but accepting the results from the app as the thing to actually start playing. As the creators, we've almost exclusively used it to choose movies or series to watch, and have often ended up watching things we'd have never found naturally in XYZ streaming service's UI (so at worst, we created an expensive away for us to avoid getting frustrated finding something we agree on).
It's limited to usage within the US only for now, as wrangling all of the metadata is time consuming (and we're based in the US). The UI/UX needs some help, as it was built by backend devs with React Native. The backend is Scala/Play/Redis/Postgres deployed on Docker Swarm.
A simple sms based chat bot for generating and mailing postcards from photos you’ve taken yourself.
I took the sms approach because I’m exhausted from always needing to install an app for everything. Most things don’t need apps. This interface is unique and kinda fun to use. Also, it took me just a weekend to whip up so I’m quite fond of it .
A cross-browser extension development framework. Got some good feedback on it on here earlier (https://news.ycombinator.com/item?id=37394194) and still hacking away at it. Appreciate any feedback!
Great projects on here so far!
LLM trained on your personal chat and texting history with a specific person, then allows you to continue talking to the AI version of them for all time.
Catnip is a result of my frustration with every other RSS reader forcing some horrible reader mode down your throat. I wanted something that was just a simple interface of aggregated links, like hacker news. So that's what you get if you'd like to try it.
A while back I really needed a way to invite some friends to a party and thought about how nice it was when we all used Facebook Events. So I made Facebook Events minus the Facebook (or any account).
You can make an event, send the link to your friends, and they can RSVP. Editing of events and RSVPs work via django sessions. No accounts or logins required.
A week after I shipped it someone showed me Partiful which is pretty much better in every way. But its still kinda regularly used among my friend group which is cool.
https://github.com/Dicklesworthstone/prepareprojectforllmpro...
https://marketplace.visualstudio.com/items?itemName=JeffreyE...
I made this because I realized I was wasting a ton of time preparing the same basic prompt format in a text editor over and over again. This extension saves me a huge amount of time. There are days when I use it 50+ times!
I built this desktop app to provide a gui for managing environment variables. I bounce between many environments and run scripts pretty regularly, so it helps!
No login, no network requests. Just uses your file system.
A collage and mood boarding website. You can browse people's boards and move cards around Front end in rust running through wasm(egui) Can drag and drop pictures and gifs, create text blocks. Also has a fun noise based background generator
The signup experience isn't very polished right now.
We should probably think about training the next LLM on all the world's magazines as well as all the books.
It is less than half-baked. I asked GPT and it said the project was "barely even kneaded", which is perfect.
I hack on it every day. I have hundreds of thousands of magazines to upload and there are many millions more hiding in obscure parts of the Internet, already scanned and waiting to be found. Plus all the amazing ones which haven't even been scanned. I intend to set up a non-profit to scan all the ones I can get my hands on.
(registration is broken, so don't try that lol)
It's a monte-carlo powered retirement simulator. I use it from time to time to do my own planning. I thought people would like it, but I think it's a little too complicated for most people.
I think it's pretty neat though - the best part is that it lets you do what-ifs and compare different scenarios, so you can see how financing that addition to your house compares to paying cash, for example.
I haven't worked on it in a while, but might circle back around to it in the next few months.
An anonymous chat app with chat rooms and an AI chatbot. I wanted a chat app that didn't require a phone number or creating an account. I don't have any plans for it anymore — I'm focusing on building https://pinjour.com.
My 13-year old son and I have been working on a behavior-oriented personal finance tool. I've kicked this idea around for years, and we had a chance to use it as our subject for a 12-week entrepreneurial program. I built a version of it for personal use years ago, but had fun looking at it through the lens of a more serious endeavor.
It's not feature complete, but the demo of the SMS-based journaling input works pretty well: https://spendlight.app/demo
A hosted sports data warehouse. Pivoting to just hosting the data rather than trying to provide a query and publish interface, which was a massive distraction because it’s so far outside my skill set.
It's reddit but all subreddits/posts/comments also get responses from unique ChatGPT personalities
This is a bit more than half-baked. I created this project because I wanted a platform for code interviews with video chat that was free and didn't require creating an account. So I decided to build it and share it for free. I still have things to do (mostly internal tools and back-end stuff) and I will keep developing it based on user feedback.
I got a lot of feedback that the sign up process is too difficult for most users. The decentralization component also makes it a bit too complex.
I am working on simplifying it extremely and also making the site extremely minimalistic. Like almost not having a sign up process at all. Would also make the decentralization part a lot simpler. Might be unpopular but I am also removing a lot of existing feature to make it as minimal as possible.
It’s a little tactical SoccerBoard to replace a whiteboard when coaching kids.
A small electron app that let me take notes while interviewing candidates and save the notes in a format. I used the rating system of chess moves (!!, !, !?, ?!, ?, ??) to mark down notable points of the interview and what i though of them. It even has a little keyboard i build with an ardunio to enter in the ratings subtly.
It was a fun project, I learned quite a bit about electron and software for embedded systems. But the hiring freezes hit and it took a bit of motivation from me as well.
Someday when things are better Ill boot it back up. and maybe even use it for interviews again.
I'm building Muziko, which is an app that will give you a list of songs every day to practice on your musical instrument of choice. I was planning to post it here eventually, but not for a long time, as I still need to build a UI that doesn't completely suck and fix some big issues. However, if this sounds cool, please have fun with it!
It's written in C++/Qt/QML. Native apps that are also completely crossplatform ftw.
Wanted to make a cartoony dark souls with a built in level editor. Games have so many fun rabbit holes that it's hard to actually make something. Then I tried to make a writeup and found out writing is hard.. still had fun!
It has one specific neat feature which is that it can pack itself into a single HTML file so you can share your mood board with someone in a very simple way.
https://thot-experiment.github.io/big_mood/
Here's a mood board for some art I'm currently working on, a single HTML file generated by clicking the "save" button (NSFW warning (artistic nudity)). https://thiic.cc/heavy/mood.html
A very rough prototype of a game library with modern functional React. After seeing only a few game libraries in React, and all being relatively old and class-based, I wanted to learn how to make a React-based version of some simple games.
So far I got pong and some walking/animating sprites on a map, but hopefully will find more time in the future to flesh it out.
One big issue I found out was it’s incredibly difficult, if not impossible, to dynamically add/remove react components in and out of the dom. React very much encourages not touching or modifying “children”, with most apis that do that being obsolete and deprecated.
The programming language project will be the next golang or rust. It has not yet released an official version.
Effindice is a browser-based passphrase generator implementing the EFF Dice method without physical dice. It is a simple javasript GUI application encapsulated within a single HTML file, designed to be used offline. It utilizes the javascript-fortuna implementation of the Fortuna pRNG algorithm and incorporates randomized delays to reduce predictability.
Hacked it together really quickly, emailed it to Bruce Schneier, not realizing he also wrote Fortuna pRNG, unexpected thumbs up
I wasn't happy with using random websites for generating passphrases like this.
I've been meaning to write a CLI version and add a feature to reorder the words into a more understandable sentence...
It’s an open source ERP system. I’ve been plugging away for about 10 months, but ERP systems are soooo big.
Really I just wanted to build graph-based, stochastic production routing, but there are so many pre-requisites to good scheduling/routing. Like the people, and the parts in stock, and when the required parts will arrive. So I’m building all that first.
It’s also been pretty enjoyable to learn Remix and Supabase in the process. I can’t say enough good about them!
As a mentor/engineering leader, I have always resorted to recording videos about how a project is structured, what decisions led to the current status and thought process behind this. I felt that much of this information is already there in the git commit log and perhaps project management software like JIRA.
gitPlay starts at the first commit instead of the head of the log. And you see a video player like interface. When you hit play, you see the folder structure evolve. I am adding file viewers, support for multiple folder browsers.
Future ideas: - Show hot spots across the entire file structure in the whole timeline - Visualize folder/file chances over time as a tree (instead of explorers at the moment) - Integrate with project management to overlay tickets with correspond to pull/merge requests - Overlay a video with a file documenting sections of the video related to commits so gitplay will show contents as you play the video
How I wish to maintain/monetize: - Keep core software open source, free for anyone, runs locally on your git repo - Pay to tap into private project management data or explainer videos (point 4 above)
Cloud music sharing for music that’s still half baked. Built this for myself years ago and I use it often when working on musical projects with friends around the world.
Think Whats app private groups meets Dropbox meets iOS voice recorder, with cloud transcoding. Mobile and web compatible.
TL;DR agi-pack is a Dockerfile generator for machine learning (ML) developers that is simple, hackable and extensible.
Fun fact: More than 75% of the original implementation was generated by conversations with GPT-4 + Github Co-pilot.
Remote build caching for everyone: https://less.build
Thanks for posting this thread :)
Godspeed is a todo app that's fast and 100% keyboard driven. I'd say it's about 3/4 baked. I use it every day, but I haven't quite gotten it over the line to a 1.0 with a payment page and all that.
http://urza.sytes.net:8080/cfusion/Enigma/Core/View/Constell...
iirc, the original ideas came from some clever team from MIT or similar setup. Used for bringing Sketchup and Blender models to life via JS. Intended for local LAN implementations.
My attempt of ms access for the web. It is version 0.8, then I hit caveats and limitations of using blazor, so i am in the process of rewriting it on asp.net core 7 mvc
tables, forms and reports are defined in xml, it has basic querying facilities and account and security management. It uses Lua for backend scripting. It allows for full app development in the browser. database is sqlite, but i made an orm that is database vendor agnostic just in case
Created an app for simple course management which is working fine, including certificate generator.
It's a (story-)scripting language based on Markdown and Lua. Basically it lets you write choice-based interactive fictions in Markdown (and Lua code blocks for advanced usages).
It is not actively developed these days since I consider the language itself is somehow usable and I have little free time these days. But the runtime-time API and compile-time macro API will probably need tons of improvements and I am still looking to implement a few features, like subroutine calls and threading support (not those threads though), and probably an IDE or at least a LSP implementation. (But again, I am quite occupied these days so the to-do list just piles up...)
Any feedback will be greatly appreciated!
Leporello.js is an interactive functional programming environment designed for pure functional subset of JavaScript. It executes code instantly as you type and displays results next to it. Leporello.js also features an omnipresent debugger. Just position your cursor on any line or select any expression, and immediately see its value. Leporello.js visualizes a dynamic call tree of your program. Thanks to the data immutability in functional programming, it allows you to navigate the call tree both forward and backward, offering a time-travel-like experience.
https://github.com/yhippa/cod-wz-2-drop-randomizer
I had so many fun ideas for this but once I got the basics down I stopped working on it further.
Let your creativity run wild as the AI generates simple, 7-page stories based on your (child's) input prompt. Practice reading by pressing the "Record" button and recording yourself reading the page, then press "Stop". StoryScroll transcribes your voice, and compares it to the original text, giving a points rating. Any mistakes are noted and added to your profile (for teacher review or further practice).
Keep a daily streak going! How long can you go for? Have a read in our virtual "library"! Earn points and dress up your virtual avatar! (Not currently available) Log in and try one of our hilarious minigames! Create a "Teacher" account and monitor your students' progress!
(Some features not currently working on iOS) Edit: Also, we turned off the image generation feature for now, because it was too expensive. It just shows a random placeholder image.
https://github.com/lukehoban/ten
Ten is a statically typed tensor programming language for defining AI models.
Ten has the following features: (1) Succint syntax and operators tailored to AI model definition (2) Fully statically typed tensors, including generic functions over tensor dimension and batch dimensions (...) (3) First-class hyper-parameters, model parameters and model arguments for explicit model specification (4) EinOps-style reshaping and reductions - tensor dimensions are explicit not implicit
Example (a functional GPT2 implementation):
Gelu(x: {...}) -> {...}:
return 0.5 \* x \* (1 + Tanh(0.7978845608 \* x + 0.044715 \* x\*3))
SoftMax[N](x: {...,N}) -> {...,N}:
exp_x = Exp(x - Max(x))
return exp_x / Sum(exp_x)
LayerNorm[S,E]|g:{E},b:{E}|(x:{S,E}) -> {S,E}:
mean = Mean(x)
variance = Var(x)
return g \* (x - mean) / Sqrt(variance + 1e-5) + b
Linear[N,K]|w:{N,K},b:{K}|(x:{...,N}) -> {...K}:
return x@w + b
FFN[S,E]|c_fc, c_proj|(x:{S,E}) -> {S,E}:
a = Gelu(Linear[E,E*4]|c_fc|(x))
return Linear[E*4,E]|c_proj|(a)
Attention[Q,K,N,V](q:{...,Q,K}, k:{...,N,K}, v:{...,N,V}, mask:{Q,N}) -> {...,Q,V}:
return Softmax[N](q @ Transpose[N,K](k) / Sqrt(K) + mask) @ v
MHA[H,S,E,K]|c_attn, c_proj|(x:{S,E}) -> {S,E}:
q, k, v = Linear[E,E*3]|c_attn|(x) {S,(3,H,K) -> 3,H,S,K}
causal_mask = (Tri[S]() - 1) * 1e10
out = Attention[S,K,S,K](q, k, v, causal_mask) {H,S,K -> S,(H,K)}
return Linear[E,E]|c_proj|(out)
Transformer[H,S,E]|mlp, attn, ln_1, ln_2|(x:{S,E}) -> {S, E}:
y = x + MHA[H,S,E,E/H]|attn|(LayerNorm[S,E]|ln_1|(x))
return y + FFN[S,E]|mlp|(LayerNorm[S,E]|ln_2|(y))
GPT2[H,S,E,B,V]|wte, wpe, blocks|(inputs:{S}) -> {S,V}:
x = wte.[inputs] + wpe.[Range[S]()]
z = for i in 0...B: x, y -> Transformer[H,S,E]|blocks.[i]|(y)
return LayerNorm[S,E]|ln_f|(z) @ Transpose[V,E](wte)
If so, I made this little app to help me quickly learn ukulele song chords nearly a decade ago: https://ukey.vercel.app/
Code: https://github.com/namuol/ukey
The idea was to translate generic guitar tabs into visual chord diagrams for ukulele simply by copy-pasting note notation. Technically it could be adapted for other fretted instruments with a few lines, because it actually generates the chords mathematically with the help of the excellent Teoria library.
This sometimes leads to questionable finger positions, but you can tap any chord diagram to see if there are alternative fingerings, which makes it easy to find the most comfortable way to transition between chords.
On a similar note, you can transpose the song until you find a key that avoids the most difficult chords.
At one point I considered automating this process by creating a simple heuristic to determine the “difficulty” of a chord fingering, but I decided this was too subjective and varies a lot depending on the adjacent chords and frequency of switching finger positions.
Needless to say, I never really learned to play ukulele…
Solution: Created a Discord Server where you need to give 3 feedbacks to request 1. This way it encourages people to participate in the discussion instead of just spamming their ideas.
Traction: 40 signups so far.
llama.cpp native chat app for macOS
runs well with metal on m1/m2, pretty slow on intel
An unfinished ASCII art point-and-click adventure game to play in your terminal.
I wanted to build a Monkey Island style game engine, but my total lack of drawing abilities only allowed me monochrome text art as a way to get some graphics.
The game engine in itself is closed to done, but I got stuck on the scenario and graphics to finish it.
Lots of rough spots. Especially the UI is not great right now but I also could not find a lot of traction so far.
Also: https://chrome.google.com/webstore/detail/sheetshortcut/fnbb...
An extension to show shortcuts as a popup for google sheets or excel 365. Never really polished it or created a landing page around it.
It's a fully client-side, Tableau-inspired data visualization & exploration tool for CSV / tabular data. I built it around 7 years ago with just ES6 and d3.js (no React etc) but never really finished or "launched" it - perfectionism led to a neverending todo list, and with a target market of only myself I didn't have a strong pull toward solving a real-world use case. I even took some time between jobs thinking I'd build it into a real product, but only managed to burn myself out instead (quite a surprise to learn that could happen on a passion project!)
But it does work - drag in a CSV file (ideally <100MB) and you can make some cool charts and quickly explore a dataset, all without the data leaving your machine. The features aren't very discoverable, but it has stuff like adding derived columns, various fancy charts, row/column sorting, custom colors, etc. I'm sure there are far better tools around these days, but I've still come back to this thing a surprising amount in my day job when I want to whip up a quick histogram or investigate some trend in a medium-size dataset.
Generator of encounter maps for tabletop RPGs. You can export to .dd2vtt format which you can then import in Foundry VTT.
There's also a new version with programmable pipeline of generators, but it's even more unfinished:
I write a lot of shell scripts, but hate how difficult it is to write things that are readable and not error-prone. I wrote a python library that allows you to add safety to any shell command while also letting you take advantage of a "real" programming language like python. Basically, this allows you to sprinkle as little (or as much) shell into your python scripts as you want.
Would love to hear any feedback on it :)
Mobile friendly meteorology data and forecasting tool.
Meant to be a better mobile option than a site like the College of DuPage model viewer.
Started it some months ago because i needed it and it didn’t exist back then. Now there are some similar competitors but it still has some unique properties.
Traction: regular signups, paying customer, some recurring users.
Needs some UX love and features/polishing to be self-servicable.
“ This app will create a Virtual MIDI device with 8 sliders and 7 buttons. Each virtual slider is controlled using a piano-key + pitch-bend gesture from your MIDI keyboard.”
My cat has been sick for a while. We have to track everything she does. My wife and I got tired of writing it on paper or a shared doc. It's also annoying to try to show the vet. So I quickly put together an mvp to track stuff and show it. It's rough and it's got only the basics. You can add a pet, events and images. (you can censor event photos, no one wants to see bloody poop pictures all the time). It's full of half done features, spelling mistakes, hacky css/html and a joke homepage lol, but it's been invaluable so far.
I use it to create symbolic links like GNU Stow but using PowerShell. This way I can keep the same folder structure inside my dotfiles repository and use Stow on WSL, and PStow on Windows.
It creates symlinks, but it can't delete them, because it doesn't keep track of what it creates. But since I don't actually have to configure too many things, that's not really a big problem for me, because can I simply delete them manually if I need. Maybe some day I will add that functionality, maybe not.
Offline WYSIWYG markdown editor + publishing tool with a Notion-like UI.
You can load a local folder from your computer by clicking the "open folder" button on the splash screen. There you can create and edit new files on your machine (all from your web browser, or install as a PWA)
The publishing feature enables you to publish a folder as an API (eg. all the data from https://codefire.dev/ is served by an API I created with mindgarden. I also use MindGarden for my blog on my personal site). Admittedly the UX to do this is not straightforward yet, I'll have to improve that and/or make a video demonstrating how to do it.
Ctrl+Shift+P opens up a go-to menu to easy navigate to any file, and Ctrl+Shift+L toggles dark mode.
Definitely half-baked and needs a lot of polish (eg. not designed for mobile yet). Let me know what you think though. May or may not motivate me to actually finish this damn project.
Written in Rust. I’m still working on it in my free time, at current pace I’ll have it feature complete by 2050.
Clippings is your modern-day archival tool for online classified ads... like Craigslist!
Explorer and visualizer for the Mandelbrot fractal.
You can customize the colors / style (lots more I want to do with this). I also have some social features in the works :) I'd like to offer the ability to order prints at some point.
Here are a few examples of cool locations you can find:
* https://fracvizzy.com/?pos[r]=-0.801141&pos[i]=-0.165043&z=1...
* https://fracvizzy.com/?pos[r]=-0.809344428&pos[i]=-0.1625402...
* https://fracvizzy.com/?pos[r]=0.321152921608&pos[i]=-0.03720...
By the way, I'm available for hire. Check my profile for deets.
Helps me schedule international meetings when writing emails.
https://chrome.google.com/webstore/detail/calzone/ahnbgncjjn...
My NAS has a load of files dumped onto it over the years. Some are organised neatly, some are in a dumping ground that I promised myself I'd sort through and never did. Some are in old folders that I haven't touched in years. Instead of spending a couple weekends sorting through it like a sane person, I'm making something to index it and hopefully replace Plex in the process.
The goal for marver is to index all those files and rip as much metadata as possible from each, then display that in a pretty UI and give the user multiple ways to refine the metadata, search through their files and make the most of the piled up 1s and 0s they have.
So far I've mostly just been having fun making individual pieces - image search powered by CLIP, extracting information from hard to parse file names using llama.cpp, pulling information from EXIF on images, some building blocks for intro detection on videos. Right now I'm porting the machine learning code to rust so I can bundle it all into a single, small docker container a lot easier, and to learn how the black boxes I'm using actually work, at least on a basic level.
I've tried similar apps but they all either have mediocre UI, are painful to setup, or make me do all the indexing manually. I want something that most people can deploy in a few minutes and that can show off a lot of what it can do immediately, while still letting the user refine the results from automated approaches. The UI has to be amazing to bundle this all into something that is actually useful, but it's been really fun putting it all together, piece by piece.
It's still a bit all over the place, but my NAS has annoyed me enough recently that I'm going to be focusing on getting a version that can be deployed, even if it's missing most of the features I want, so that it can start being useful to me. I think once that happens I'll be more motivated to add features, especially with a solid base I can work from.
It started as a wired replacement for wireless communications by emulating an ANT+ USB stick; the host software thinks the packets are being sent wirelessly.
Then it become more of an exploration into how inertial trainers calculate input torque only from the known properties of the flywheel. In theory it could become an open firmware for any trainer, since the hardware is largely the same. I never really finished tuning the functions to estimate the power from the flywheel RPM.
I’m only a few weeks in, but Insomnia’s complete f*ckup just days after I started really proved I’m on the right track!
Hoping to get a simple site setup next week, the day job has been exhausting recently…
Using the browser tools is normally better though (change everything, to anything), but I think a visual interface with sliders beats out entering hex codes for colors.
I don't know if I'll ever get around to adding any new features, but it's complete in that it works and does basically everything the old UI did.
A bicycle computer. Hackable. Open source software. Off the shelf hardware.
ARM microcontroller, 100% Rust, Bluetooth (not yet).
I have had a bike computer nearly 3 decades ago, but the dream to have the distance counter reset automatically has remained unfulfilled. I could either get a fixed-function calculator or 100% closed overpowered gadget. Why can't I have some fun myself?!?
Thankfully, now we have cheap, reverse-engineered smart watches. I found a pretty decent, sunlight-readable one, and now I'm hacking away!
The big TODO is speeding up development by switching from TockOS to RTIC, and implementing a minimal Bluetooth Low Enegy stack.
Currently on hiatus because it's bike season and I'd rather spend my time outside :)
Generate React code from any design system with a simple text prompt. Trying to figure out if the code output is more useful or if the design "inspiration" is more useful. Would appreciate any feedback!
It's a search engine / social network prototype I've been working on.
My idea was to use a social network as the backbone of a search engine.
My thinking was that the majority of the internet doesn't need to be indexed, so why not just use the portion of the web that is offered up by users? If this sounds stupid, I would offer Hacker News as a working example of this model.
My other motivation was that the majority of platforms operate on a popular vote strategy, as a result, content that appeals to our lesser nature tends to do very well. YouTube, for example, could not support the development of content as ambitious as Avatar.
Content on Zohing is ranked using a "recursive voting" strategy. It attempts to find accounts that have been deemed by other users to be "high quality" and weight their interactions more.
Put simply: On the spectrum of Avatar to TLC I want creators to aim for Avatar. I am try to create a platform whose underlying game incentivizes the creation of difficult, high effort content instead of reality TV and cable news.
Zohing was created to function as a mobile web app. Getting it to work smoothly on iOS safari was a real challenge. I had to do some ugly hacks to get 2 way infinite scrolling to work. For version 2 I am going to go with a more conservative approach.
If you want to know how the recursive vote algorithm works or have any other questions you can email me at: surgeon@zohing.com
I also have Free Hero Mesh, which is good enough to be used now although it lacks installers, and the puzzle catalog service, etc; and is meant only on GNU/Linux; so you would have to compile and set it up by yourself, although hopefully people who are able to port it, and other help with it. (Also a few features are missing; e.g. currently font sizes are limited to 8x8, and there is no music (although sound effects are implemented).)
Reddit reimagined based on Personality
A few years ago I had the idea for a service that would allow you to better understand what type of coaching and feedback managers were giving staff. Still in beta, but there are a few ways to make this service more useful for first line and functional managers.
Why mRuby? Well this is the first step of a larger dumb project where I'd have pure Ruby as Linux init/userland and boot straight to it (like these old Amstrad or whatever booted straight to BASIC), then maybe even write a toy kernel in Ruby and drop Linux as well.