Of the ten postings, Nines (ID: 22672013) is the most revealing because of what it doesn't say. There's no stack list, no framework name-drop, no "we use Kubernetes and Rust." Instead, the entire posting is a founder-and-investor pedigree pitch: David Stavens (ex-Udacity, Stanford self-driving), Accel, 8VC. That absence is the signal.

1. The implied stack. Radiology AI in 2020 means a PyTorch/TensorFlow imaging pipeline, DICOM ingest, GPU inference clusters, and a HIPAA-compliant cloud (almost certainly AWS with BAAs). The fact that they don't bother listing it suggests they're hiring for judgment, not framework familiarity — they want people who can architect a regulated medical imaging system from scratch, not plug into an existing one.

2. Stage and direction. "First-of-its-kind radiology service" combined with "our radiologists and engineers" tells you Nines is vertically integrated — they employ the radiologists, not just sell software to them. That's a much harder business than a pure SaaS play: they're competing with teleradiology incumbents (Nighthawk, vRad) on service quality while simultaneously building the tooling. The Stavens/Waymo lineage hints they intend to apply ML to reads, but they're carefully not promising "AI radiologist" — likely because FDA clearance for autonomous diagnosis is still a multi-year slog.

3. Skills and trends highlighted.

• Vertical integration as a strategy — owning the service layer to generate proprietary training data is the 2020 ML moat playbook.
• Founder-led recruiting — the entire pitch is "come work with Stavens," which works when you're pre-Series-B and selling on optionality.
• Generalists over specialists — "Software Engineers" with no level, no specialization, no stack. They want builders.

4. Flags.

• Green: Iconic investors with healthcare exits (Oscar via 8VC), credible technical founder, mission with obvious patient impact.
• Green: Palo Alto onsite during a thread where most COVID-era postings are scrambling to claim "remote until post-COVID" — confidence in their working model.
• Yellow: No salary band, no team size, no engineering blog link. The pitch leans almost entirely on pedigree, which works for the first 10 hires but stops scaling around #30.
• Yellow: "Best radiologists deliver the best patient care" is a thin product thesis — what's the actual wedge versus existing teleradiology?