I've been building a board that pulls AI jobs straight from company career pages rather than reposting aggregator listings, and the category split genuinely surprised me.
Out of 18,385 live roles today: Engineering 43.5%, Sales & GTM 17.8%, Operations 11.6%, Research 5.6%, Marketing 4.7%, Product 3.6%, Policy & Safety 1.0%, and Other 12.3%.
Being upfront about that last bucket, because it's the honest caveat: "Other" is roles my classifier couldn't confidently label, and spot-checking it, it does contain real engineering titles. So "56% is not engineering" would be overclaiming. What I can defend is that 44.3% of the board sits in a named non-engineering function, which is still a lot more than the market talks about. Sales & GTM on its own is bigger than Research, Marketing and Product combined.
That matters because almost every AI job board I've used is engineering-only. A large share of people trying to get into AI are searching on boards that structurally ignore their function.