AI job scrape: evals mentions jump from 20% to 25% as integrator skills hit 91% of postings
Al_Grigor · x · 2026-10-14
Based on monthly scrapes of AI engineering job postings across Berlin, Amsterdam, London, New York, LA and India, the author maps what the market actually wants:
- Demand is up: postings grew from 895 in February to 1222 in August and 1086 in September.
- Integrators up, trainers down: the 'integrator stack' (RAG, agents, APIs) appears in 91% of postings, up from 38%, while the 'trainer stack' (PyTorch, fine-tuning, CUDA) fell to 30%; inference/serving skills mentions dropped from 20% to 16%.
- Evals are the differentiator: explicit LLM-evaluation requirements rose from 20% in February to 25% in September — the skill the author says matters most.
- Advice: focus on RAG, agents and evals, learn by building a project with real users, and skip model training/serving for now — knowing how to call an LLM API is enough to start.
More from Companies & People
- Perplexity CEO: Over 50% of agent-paying users are American, US alone can build a trillion-dollar company — AravSrinivas · 2026-10-15
- WSJ: OpenAI's computing chief to depart — arthurcolle · 2026-10-15
- AI-in-biology startup hiring MTS to build rigorous benchmarks for AI in biology — kenbwork · 2026-10-15
- OpenAI UK filings show $3.1M average compensation per employee in 2025 — emeka_boris · 2026-10-15
- SF Tech Week hits 53,991 attendees but no-shows spike as AI signs people up — FinanceYF5 · 2026-10-15
- Jensen Huang Says He Won't Sell GPUs to Irresponsible AI Labs — rohanpaul_ai · 2026-10-15