About the role
About Halluminate
Halluminate is an applied AI and data company building computer-use and tool-use RL gyms to train AI agents on real financial services work — Excel modeling, pitch decks, and everything investment banking, private equity, and hedge funds rely on. We work with frontier labs, graduated from Y Combinator (S25), and raised a seed round to scale the team.
The role
Base: $200–275k · Equity: 0.15–0.30% · Location: San Francisco, 5 days in office
Own platform engineering, develop frontier long-horizon RL environments, and help build our engineering org from the ground up.
Responsibilities
-
Build our RL environment training and inference infrastructure to support customer usage at scale.
-
Research and develop next-generation RL environments — increasingly realistic, long-horizon, and difficult for frontier models.
-
Build software that 10–100x's the quality and throughput of RL environment creation.
-
Innovate on synthetic data pipelines to create realistic problems.
-
Build platform analytics for environment cost, hours, bottlenecks, and SME management.
-
Craft domain-specific verifiers (verifiable rewards for decks, Excel modeling, quantitative trading, and beyond).
-
Establish engineering culture and practices from the ground up.
You should have
-
Familiarity building evaluations, benchmarks, or RL for AI agents.
-
Startup speed: iterate quick, ask quick, respond quick, make mistakes quick.
-
Strong product and user ownership; ability to prioritize across a large roadmap.
