Andrew Carr backed the advice, citing paper reimplementation tests.
Don’t get degrees man. Just implement papers, tweet about it, write good blog posts, create nice repos. Biglabs will come and beg you to work for them.
I feel kinda dumb for thinking it was too late to get a phd in ML in 2018
Users are supporting the push to prioritize building projects and reimplementation skills over PhDs for AI lab jobs because the approach better reflects current hiring realities.
No Digg Deeper questions have been answered for this story yet.
Number one skill we screened for back in the day was ability to reimplement papers
Don’t get degrees man. Just implement papers, tweet about it, write good blog posts, create nice repos. Biglabs will come and beg you to work for them.
@recurseparadox isn't phd path is still the most common way to get into biglabs?
@recurseparadox Does it really work in 2026 ?
@rathore_viks Not really no
@punitvara Works better now
@recurseparadox ❤️
@recurseparadox Is that what you did?
@recurseparadox @rathore_viks yes it is
@recurseparadox Hey, can I DM you?
@recurseparadox I have my independent open source lab which does this - 350+ people doing this. Idea is to make rebel neolab