Our client is a Y Combinator-backed, seed-stage AI data research lab building the data layer for reinforcement learning. They build synthetic data pipelines and simulation environments that enable frontier AI models to learn and reason. Instead of writing traditional application PRDs or managing feature backlogs, you will act as the PM of the model itself—co-designing capabilities alongside elite AI researchers.
Key Responsibilities
Co-design model capabilities alongside researchers and diagnose performance gaps in long-horizon coding, cybersecurity, and ML chip design.
Architect synthetic data pipelines and simulation environments using our client's proprietary engine to address model weaknesses.
Execute experiments and benchmarks using internal simulation stacks and open-source tooling.
Engage directly with researchers at top AI labs and Fortune 500 enterprises to identify where state-of-the-art models fail in real-world scenarios.
Influence the product roadmap, go-to-market strategy, and overall research direction based on evaluation metrics.
Requirements
Hands-on project experience with RL training data, synthetic data generation, or simulation environments.
Proven ability to diagnose model failure modes and build targeted training data solutions.
Strong technical fundamentals (CS degree, ML background, or equivalent) with the ability to read code and understand model architecture.
Direct experience running ML experiments and interpreting benchmarking data.
High drive and adaptability suited for an early-stage, fast-moving meritocracy.