Research Engineer - Contextual Bandits & RL
Core
Build decision-making models for in-store hyper-personalization using offline contextual bandits and reinforcement learning, learning from logged human interaction data.
Role type
Research Engineer (Offline Contextual Bandits & RL)
Builds
Decision-making models for single-step and multi-step customer journeys in retail
Domain
Retail / Hyper-personalization / Offline Reinforcement Learning
Deliverable
production ML models
Required skills
Contextual bandits, Reinforcement learning, Counterfactual learning, Transformers, Graph Neural Networks (GNNs), Python, Off-policy evaluation (OPE), Dataset design, Production-level code debugging
Preferred skills
Offline policy learning and evaluation methods (IPS, doubly-robust), Bandit algorithms and exploration strategies, Recommender systems and ranking, Data pipeline construction
Technologies
Python, Transformers, GNNs
Responsibilities
Develop and productionize offline contextual bandit and offline RL methods; Build rigorous off-policy evaluation and counterfactual validation; Formulate single-step and multi-step decision processes based on real retail interactions; Advance representation learning for decision-making; Translate research ideas into robust systems including deployment and monitoring; Collaborate cross-functionally to turn ambiguous product goals into concrete ML objectives
Seniority
Mid-Senior, hands-on IC
