Member of Technical Staff
Halluminate
Member of Technical Staff
Halluminate
Data research lab creating RL benchmarks
Halluminate is a data research lab building benchmarks and RL environments to push the frontier of AI in knowledge work.
Our first area of focus is economically valuable workflows in financial services and consulting, but we are rapidly expanding into adjacent verticals.
Base: $250–320k · Equity: 0.20–0.35% · Location: San Francisco, in-person
Lead our in-house research and post-training efforts.
Train open-source frontier models on our in-house environments to validate quality and build research insights.
Lead our public-facing benchmark and leaderboard efforts for frontier models.
Publish findings in blogs and papers to share with the broader research community.
Contribute to core platform engineering as needed — AI tooling, data collection pipelines, partnering with environment and reward designers.
Build our engineering team culture from the ground up.
Experience in model post-training.
Experience publishing papers and sharing open research.
Familiarity building evaluations, benchmarks, or RL for AI agents.
Startup speed: iterate quick, ask quick, respond quick.
Strong product and user ownership.
Client-facing comfort — you'll talk to users, customers, and SMEs.
Former founders or experience at early-stage startups.
Healthcare — Fully covered medical, dental, and vision for you. Subsidized for dependents.
Relocation — $10k+ relocation stipend to help you get to San Francisco.
401(k) — Up to 4% employer match.
Food — Free breakfast, lunch, and dinner while working.
Health — Free gym membership.
Transportation — Free Ubers and Waymo rides to and from the office.
Visa sponsorship — We sponsor visas, including H-1B.
Data research lab creating RL benchmarks
Halluminate builds benchmarks and reinforcement learning environments to advance AI for knowledge work. It creates computer-use and tool-use environments that train frontier AI models on real finance and consulting tasks, such as investment banking, private equity, and consulting work. A flagship benchmark, Westworld Finance Diligence Bench, contains 88 problems that simulate a complete company acquisition due-diligence process, with content written and reviewed by practicing deal professionals. The company differentiates itself by delivering domain-specific, practitioner-authored tasks and end-to-end workflows across financial services and consulting, and by expanding into adjacent verticals through data-driven RL environments. Its goal is to raise AI performance on real-world knowledge-work tasks and broaden benchmarks to cover more industries, supported by a 2026 Series A funding round led by Oak HC/FT and investors like Y Combinator.
Explore more opportunities at Halluminate