/roles — ROLE_330
Member of Technical Staff - Platform Engineering
Fast-growing startup building long-horizon RL environments and benchmarks for training AI agents in finance
The role
- COMP
- $200K - $250K
- EQUITY
- 0.15 - 0.3%
- LOCATION
- San Francisco
- WORKPLACE
- On-site
- EXPERIENCE
- 5 - 12 years
- VISA
- None, Visa transfers
- STACK
- Python, TypeScript, Docker, AWS, OpenAI
- INDUSTRY
- AI, Data, Finance, Financial Services, Fintech, Insurance
The company
Small, fast-growing startup building realistic reinforcement learning environments and benchmarks that frontier AI labs use to train agents on real financial and knowledge work.
- STAGE
- Series A-stage
- FUNDING
- $8M raised
- TEAM
- ~10 people
- FOUNDED
- 2024
- BACKING
- VC-backed
JD — the work
About the role
As a founding member of technical staff, you would lead platform engineering, build long-horizon RL environments for frontier models, and help create the engineering organization from scratch. The environments target financial services work such as spreadsheet modeling, slide building, and trading, and they need to be realistic and hard for today's best models. It is an in-person San Francisco role with plenty of customer contact.
What you'll do
- Build the training and inference infrastructure behind RL environments so customers can use them at scale
- Investigate and create more realistic, longer-horizon, and harder environments for frontier models
- Build tooling that raises environment-creation throughput and quality by 10x or more
- Improve synthetic data pipelines that generate realistic problems
- Create analytics that track spend, time, bottlenecks, and expert contributors across environments
- Design verifiable reward checks for finance tasks like slide decks, spreadsheet models, and quant trading
- Shape how the engineering team works, its norms and habits, from day one
What they're looking for
- 5 or more years of hands-on software engineering experience
- Hands-on work on evaluations, benchmarks, or reinforcement learning setups for agents
- Startup speed: fast iteration, quick questions, quick replies, and fast learning from mistakes
- Product ownership and the judgment to prioritize a long feature roadmap
- Comfort working directly with customers, users, and subject matter experts
Nice to have
- Research experience such as training models or publishing papers
- Previous founder or early-stage startup experience