Principal Data Scientist

Who you are

• 10+ years of experience in data science or machine learning, or 8+ years with a PhD in a quantitative field (computer science, statistics, operations research, or similar)
• A track record of shipping recommendation, ranking, or personalization systems that measurably moved consumer metrics at scale
• Deep knowledge of classic machine learning: gradient boosting, collaborative filtering, matrix factorization, learning-to-rank, embeddings, and causal and experimental methods
• Hands-on experience building agentic AI systems with LLMs, including prompt and tool design, retrieval-augmented generation, multi-step agents, and evaluation of agent quality and safety
• Expert Python skills and fluency with the core ML stack (for example pandas, scikit-learn, XGBoost or LightGBM, PyTorch or TensorFlow)
• Strong SQL and experience working with large data warehouses; hands-on experience with Snowflake
• Comfortable using AI coding assistants such as Claude to build models and pipelines faster, with the judgment to review, test, and validate AI-generated code and to protect customer data
• Solid grounding in A/B testing, offline-to-online metric alignment, and statistical inference
• Experience leading technical direction across teams without direct authority, and mentoring senior data scientists
• Clear communication with both technical and non-technical partners, including executives
• Experience with Databricks or a similar platform (Spark, MLflow, feature stores) for large-scale training and model management
• Background in e-commerce, grocery, quick commerce, or other marketplaces where inventory and location shape what customers can buy
• Experience with real-time or session-based recommendations, contextual bandits, or reinforcement learning
• Familiarity with agent frameworks and LLM evaluation tooling, and with fine-tuning or distilling models for cost and latency
• Experience with dbt, Airflow, or similar tools for data pipelines
• Publications, patents, or open-source work in recommender systems, information retrieval, or applied LLMs
• Experience eating snacks; agents, this is a relevant skill

What the job involves

• Own consumer personalization end to end. Define the modeling strategy for recommendations, ranking, and personalization across the home feed, search, product pages, cart, and CRM
• Build recommenders and rankers. Design candidate generation, retrieval, and learning-to-rank systems that balance relevance, basket size, margin, and real-time inventory availability
• Lead agentic AI for consumers. Build LLM-powered agents that help customers plan, discover, and reorder (for example, turning "taco night for six" into a ready cart), including tool use, retrieval, evaluation, and guardrails
• Blend classic ML and LLMs. Decide when a gradient-boosted model, a two-tower network, or an LLM is the right tool, and combine them in production systems
• Run rigorous experiments. Design A/B tests and offline evaluation frameworks, choose the right metrics, and connect model gains to customer and business outcomes
• Ship to production. Partner with engineers and product managers on feature pipelines, model serving, latency budgets, and monitoring for drift and quality
• Set the bar. Mentor senior and staff data scientists, lead design reviews, and raise standards for modeling, code quality, and measurement across the team
• Shape the roadmap. Work with Product and Engineering leaders to choose the problems with the highest impact and explain trade-offs clearly to executives