Lead Data Scientist, Demand

About this Role: Every day we decide how much fresh food to send to each fridge and retail store. Send too little and we lose sales and customers. Send too much and food expires. Our demand planning models sit at the center of that call. The Lead Data Scientist, Demand owns those models. You'll be our first full-time data scientist dedicated to demand planning. We have fridge and retail forecasting models in production today. Your job is to maintain them, make them better, and rebuild them where a better approach pays off. As the core models mature, you'll expand into company-level volume and revenue forecasting. You'll work closely with our Lead Decision Scientist. You own the predictive layer: what will sell, where, and when. They own the prescriptive layer that turns your forecasts into stocking, allocation, and routing decisions. You'll also partner with Data Engineering to run your models in production, and with the demand planning team who use your forecasts every day. Your day-to-day partner is our Manager, Demand Planning. Channel Strategy and Demand Planning set the business rules and objectives your models run on: availability targets, the trade-off between lost sales and waste, and the operating constraints planners live with. Your job is to turn that logic into models that follow it, and to tell us when the data says a rule is wrong. You set the technical bar. The Channel team sets the commercial one. What You'll Do... - Own our fridge and retail demand forecasts end to end: data, features, model, validation, and production. - Build forecasts that maximize contribution margin dollars, not forecast accuracy for its own sake. Track the impact through sell-through, unsolds and stockouts against Channel's targets. Watch for forecasts that run consistently high or low by item and fridge, and give the optimization models a realistic range to achieve business goals. - Maintain and improve the models in production today. Rebuild them where a better approach clearly pays off. - Forecast at the item, location, and day level across 2,000+ fridges and retail partner stores, where demand is often low-volume and intermittent. - Model the hard parts of fresh food: stockouts that hide true demand, short shelf life, new fridges ramping up, new menu items, holidays, and promotions. - Produce probabilistic forecasts, not just point in time estimates, so our optimization models can weigh the cost of a lost sale against the cost of waste. - Partner with the Lead Decision Scientist on the forecast outputs the solver needs. Measure how forecast changes move stocking decisions and results. - Build backtesting and monitoring. Judge accuracy on business outcomes like sell-through, waste, and lost sales, not just statistical error. - Run a standing model scorecard, reviewed with Channel Strategy and Demand Planning on a set cadence. Every material model change ships with a measured before and after on sell-through, unsolds and stockouts, translated into contribution margin per fridge per week. - Partner with Data Engineering to run models in our production pipelines (Snowflake, dbt, Dagster) without manual babysitting. - Work with demand planning, supply chain, and operations. Explain what the model expects, why, and how much to trust it. - Measure the value of planner overrides. Fold the ones that consistently help into the model as explicit rules, and retire the ones that don't. - Expand into company-level volume and revenue forecasting as the core models mature. - Keep a living record of every business rule in the models: what it does, where it came from, who owns it, and when it last changed. When Channel changes a requirement, we should be able to find it, change it and see the effect in days. - Set the bar for how forecasting models get built, tested, documented, and shipped at FF. Who You Are... - 4-8 years building and maintaining forecasting or predictive models in production. - Deep demand forecasting experience: time series, hierarchical, or ML-based methods (gradient boosting, state space, Bayesian, or similar). - Comfort with count data and intermittent demand (Poisson, negative binomial, or similar). - Experience correcting for censored demand from stockouts, or a strong grasp of why it matters. - Strong Python and SQL. Experience with Snowflake or a similar cloud warehouse. - Rigor in backtesting and experiment design. You know how to prove a model change actually helped. - Track record of owning model work end to end, from problem framing through production and monitoring. - You can point to a model change you shipped and the business result it moved, in dollars. - Clear communicator. Able to explain forecast uncertainty and tradeoffs to planners and operators, and defend modeling choices to technical peers. - Comfortable in a fast-paced environment where you set direction, not just follow it. - Familiarity with using AI tools in your day-to-day workflow. - Based in Chicago or willing to relocate. This is an in-office role. Nice to Have... - Forecasting experience with perishable goods, grocery, CPG, food service, or multi-location retail. - Exposure to how forecasts feed optimization models for inventory, allocation, or routing. - Causal inference experience (difference-in-differences, matched controls) for measuring pricing, promotion, or operational changes. The base salary range for this role is $160,000 - $170,000. The base pay offered will be determined by factors such as experience, skills, training, certifications, education, and any applicable minimum wage requirements. In addition to base salary, this position is eligible for company performance-based bonuses and equity. We provide a comprehensive benefits package, including: - Medical, dental, and vision insurance (multiple plans available) - 401(k) with immediate employer match vesting - Paid time off (including vacation, sick leave, and holidays) - Paid sabbatical after 5 years of service - Employee discounts - Employee Assistance Program (EAP)