platoseedplatoseed
Helping your company find 10x engineers.
Saffron allows companies to quickly evaluate how well software engineers use AI to code. The way engineers write code has fundamentally changed, but the way companies evaluate them hasn't. Saffron gives teams full visibility into how candidates use AI agents to build, debug, and ship real features on an actual codebase. Saffron was founded by Robert Chondro (formerly MIT and Jane Street), Jerry Yao (Stanford and Jane Street), and Kazuma Choji (Harvey Mudd, published in NeurIPS and ICML), and is backed by Y Combinator (P26) and Afore Capital.
Saffron provides AI-native technical assessments that show how candidates actually build with AI, including session replay, AI reliance scoring, and multi-agent code review. The platform targets engineering hiring by evaluating real code from candidates against custom rubrics.
Saffron connects to your codebase or templates, creates assessments with 10+ AI review agents and custom evaluation criteria, invites candidates via a single link, and records full session data. Candidates build on your real repository using a browser-based IDE with Claude Code, and Saffron scores each submission with multiple AI agents, distinguishes human-written vs AI-generated code, and offers session replay and AI-checked debrief questions. The platform enables comparison to a traditional interview loop and provides various pricing tiers with assessment limits, code attribution, and replay capabilities.
Who it’s for: Engineering teams and talent acquisition teams at mid-market to enterprise companies seeking to assess candidates’ real coding ability and AI utilization in production-style work.
Funding mentioned (Y Combinator); productized pricing tiers and enterprise options; focus on hiring efficiency and AI-enabled assessments
CEO Studied math & computer science @ MIT Building AI-native technical assessments to vet/hire SWEs
Building software to test how well swe's code with ai _______ previously: cs/math @ Harvey Mudd High school NeurIPS, ICML, and ICCV author
Building AI-native technical assessments Prev. CS + math @ Stanford
Companies use Saffron to evaluate how well software engineers use AI to code.
Saffron gives engineering candidates a real codebase and access to AI tools to complete a task while capturing prompts, diffs, and decisions in a sandboxed environment. The platform automatically scores how candidates use AI and build code, not just whether output compiles.
▲ 33
The AI-native observability platform for fast-moving engineering…
Hire engineers based on what they’ve shipped
