Principal Applied Scientist (User Modeling)
Core
Building user world models and LLM post-training pipelines (SFT, DPO, RFT) to power agentic AI for home shopping experiences.
Role type
Principal Applied Scientist (User Modeling & RL)
Builds
User world models, reward models, and on-policy assessment frameworks for agentic agents.
Domain
Real Estate / Agentic AI / Reinforcement Learning
Deliverable
production ML models
Required skills
Reinforcement learning, causal inference, sequence modeling, user behavior modeling, Python, PyTorch/TensorFlow
Preferred skills
Economics/econometrics, generative sequence models, rare-event modeling, two-sided marketplaces
Responsibilities
Own LLM post-training pipelines (SFT, DPO, RFT); Build multi-category reward models; Design on-policy and online assessment; Translate offline evaluation rubrics into reward functions; Provide technical leadership and mentorship.
Seniority
Principal, hands-on IC with technical leadership
