About the role
You work on knowledge, world models, memory, and the evaluations that keep them honest, in a lab whose partners are Stony Brook, Purdue, and the IITs. Our constraint shapes the research: the model often runs on the person's own hardware under a consent boundary, which closes off the easy answers.
What we need to see
- Research contribution you can point at: publications, or a shipped system whose core idea was yours
- You implement your own work to a standard that could ship
- Rigor about evaluation, especially at telling real gains from better-looking benchmarks
- Depth in knowledge representation, memory, world models, or retrieval
Nice to have
- Personalisation or user-specific models
- Privacy-preserving machine learning
- Open-source or open-weights contributions
What winning looks like
- Benchmark/eval gains that reach the product
- Research shipped to users
- Reproducibility and quality of work
Where and how we work
In the office together five days a week, in any of these cities. Remote-friendly around your family, arranged one person at a time.