Want to know if this job is worth applying to?
United States
Remote only
Gopuff delivers everyday essentials in minutes from our own network of micro-fulfillment centers. Every session, a customer sees a small, fast-changing assortment that depends on where they are, what's in stock, and what they need right now. Getting that experience right is one of our biggest levers for growth.
As Principal Data Scientist, Consumer, you will be the technical lead for how Gopuff personalizes the shopping experience. You will design and ship the recommendation, ranking, and personalization models behind search, browse, carts, and marketing, and you will lead our work on agentic AI experiences for consumers. You will set technical direction, mentor data scientists, and partner closely with Product, Engineering, and Marketing leaders.
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.
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.
