Principal Applied Scientist, User Modeling

What you'll do

• Own LLM post-training pipelines including SFT, DPO, and reinforcement fine-tuning (RFT/GRPO) on GPU infrastructure.
• Build and train multi-category reward models (PRMs) that emit scalar signals from contrastive preference pairs.
• Design on-policy and online assessment frameworks to score live agent trajectories using LLM-as-a-Judge.
• Translate offline evaluation rubrics into generalizable reward functions for production traces.
• Build user world models to understand shopper journeys, intent, and the impact of agent actions.
• Apply sequence modeling, causal inference, and reinforcement learning to drive production impact.
• Provide technical leadership and mentorship to scientists and machine learning engineers.

What we're looking for

• Must have a PhD in Computer Science, Statistics, Economics, Machine Learning, or a related quantitative field.
• Must have depth in reinforcement learning or causal inference.
• Must have hands-on experience modeling user behavior at scale using sequence models over event streams, sessions, or trajectories.
• Must have practical depth in causal inference or reinforcement learning applied to sequential decision problems.
• Must be able to independently scope and solve ambiguous problems while providing technical leadership to scientists and MLEs.
• Must have working proficiency in Python and ML frameworks such as PyTorch or TensorFlow.
• Background in economics or econometrics, such as demand modeling or consumer choice (preferred).
• Published work in RL, causal inference, survival analysis, or user modeling/simulation (preferred).