Do you need to run Linux natively, and/or need built-in 4G/5G WWAN, and/or have above-average RAM requirements? Then, get a ThinkPad X, T or P-series (the latter for up to 128 GB of memory).
Is Visual Studio (or any other truly Windows-only app) a requirement for your job, and/or do you want a true Linux shell experience without the hassle of native Linux? Then, get any Zen4-or-i7 / 8-or-16GB / 1TB+ NVMe Windows 11 laptop from Lenovo, HP or whomever. A clean-wipe-the-OEM-crap-away Win11 Enterprise + Office install (which your workplace most likely will have available, or you can create yourself as part of an Office365 subscription plus a bit of work) will do just fine. That plus WSL2 is a thing of, almost, beauty.
Otherwise, get an M3 MacBook. Air if it works for you (no fan! no noise!), 'Pro' otherwise. Crisp display (but heed the limitations on external monitors!), excellent performance, slightly-weird-yet-tolerable OS, not great for Linux-shell-things (I mean, Homebrew is just excentric...), but a great default choice.
However, do keep in mind that laptops will always have a limited thermal envelope, and that if you just need to wring every last bit of performance out of hardware, offloading at least a fraction of your workload to a properly-preferably-water-cooled desktop and/or server is inevitable, and may fundamentally change your requirements for said laptop...
It's repairable, upgradable (MB swap for processors (3 generations of Intel and one AMD so far), ram and storage are socketed), linux support is fairly good with most things just working (running NixOS personally on my 12th Gen Intel 13").
Performance is average (but the i5 I got was a improvement over my 2012 Retina MB Pro) and battery life is average to poor (but I am plugged in most of the time), size and weight are okay (for me!).
On the other hand, I've had good luck with Dell, Alienware and Lenovo machines too and personally prefer both Windows and Linux to MacOS. One advantage is that machines like this can have a powerful NVIDIA GPU which at the forefront of AI model training, though Apple is doing a lot to make their platform usable for that purpose too, particularly the unified memory model could let their machines hold bigger models even if they don't run quite as fast.
If you really want to "think different" get a full-sized desktop machine as this can be much more powerful.
See also the discussions here: https://tildes.net/search?q=laptop