Tecton's posting is the most revealing on the list because it's a rare artifact: the founding team of Uber's Michelangelo ML platform commercializing the exact system they built in-house. The pitch isn't "we have an idea" — it's "we already built this at scale for one of the hardest ML workloads on earth, and now we're selling it."

Stack signals from the role mix:

• UI and Data-Viz roles alongside Backend and Data Infra — this is not a pure infra play. They're building a product surface, which means data scientists (not just platform engineers) are the intended buyer.
• Data Infra as a distinct discipline from Backend signals a serious feature store / streaming pipeline underneath — the Michelangelo lineage points to Spark, Kafka, and online/offline feature parity as the hard problem.
• Two-coast onsite footprint (SF + NYC) with no remote option in early 2020 — deliberate, not accidental. NYC = finance customers, SF = tech customers.

What the posting reveals about stage: "Well funded by top-tier VCs, paying enterprise customers, excellent engineering teams" is Series A/B language — past product-market-fit anxiety, into scale-the-team mode. The name-drop of Michelangelo is doing enormous fundraising and recruiting work: it substitutes for years of credibility-building. They're essentially arbitraging Uber's engineering brand.

The trend it highlights: The MLOps productization wave. In 2018–2019, every large tech company built an internal ML platform (Michelangelo, FBLearner, TFX). By 2020, the "productize the internal platform" companies are emerging — Tecton (Michelangelo), Determined AI, Weights & Biases. Feature stores specifically are becoming the wedge because online/offline feature skew is the #1 cause of ML models failing in production, and nobody wants to build that plumbing twice.

Green flags: Named founders with verifiable pedigree (linked blog post), stated customers, no equity-vs-salary hand-waving, hiring across the stack (indicates real product, not just a demo).

Yellow flags: "Onsite" with mandatory bicoastal presence limits the talent pool severely — a bet that senior ML infra engineers will relocate. The posting is thin on specifics about the actual tech (no mention of languages, cloud, or open-source strategy), suggesting they're guarding their moat or still deciding.