Of the ten postings, Kensho's is the most strategically revealing — not because of what it says, but because of what it carefully doesn't.

The stack tell: Note the conspicuous absence of any tech stack. No Go, no React, no Kubernetes — while every other posting in this thread leads with frameworks, Kensho leads with capabilities: "deep learning speech recognition," "entity recognition," "state-of-the-art search," "AI-driven research platform." This is a tell. Companies that lead with tools are hiring builders for known problems. Companies that lead with capabilities are hiring researchers for problems they're still defining. Kensho is the latter.

What the geography reveals: "Cambridge/Boston, NYC, DC, LA" — four offices for a single role posting is unusual. This isn't a startup hiring footprint; it's an acquired-subsidiary footprint. The phrase "S&P Global's world-class data" gives it away: Kensho was acquired by S&P Global in 2018 for ~$550M, and these offices map to S&P's existing presence (NYC HQ, DC for government data, LA for entertainment ratings). They're embedding ML talent into S&P's existing data fiefdoms rather than centralizing in one lab.

The "across the stack" phrase: Hiring "Data Scientists/Engineers across the stack" with no role differentiation suggests they're still figuring out team shape. A mature ML org posts for specific roles: NLP Research Scientist, ML Platform Engineer, MLOps. Kensho is casting a wide net — they want the people first, then assign the problems. This is either confidence (we'll find work for any strong ML hire) or drift (we don't know what we need).

Green flags:

• Real proprietary data moat (S&P's financial data is genuinely irreplaceable)
• Multiple ML application areas listed — suggests actual production systems, not vapor
• The "drive fact-based, objective decision making" framing implies their customers are S&P's existing institutional clients, meaning revenue is real

Red flags:

• No salary range, no team size, no engineering blog link — opacity that's typical of finance-adjacent firms but unusual on HN
• Username kenshotech19 looks like a recruiter throwaway, not an engineer
• Buzzword density ("leading," "world class," "state-of-the-art," "innovative") in two sentences is a marketing-written posting, not an engineer-written one — a contrast to ngrok's posting where the Director of Engineering wrote it himself

The trend: Kensho represents the post-acquisition "AI lab inside a data incumbent" pattern that defined 2018-2020 — Bloomberg, Reuters, S&P all racing to apply ML to data they already owned, hiring across multiple cities to be near customers rather than near talent.