As the Recommendations Engineer at Mill, you'll own the recommendation system end-to-end — from the signals and models that decide what to recommend, to the feedback loop that tells you whether it worked. You'll be part of the Data team that also manages Data Platform, Integrations and Warehouse. You'll partner with product and engineering teams to make sure recommendations are useful, accurate, and get better over time.

What You'll Do

• Build and operate the customer-facing recommendation engine that turns food waste data into actionable recommendations — purchasing suggestions, anomaly explanations, operational nudges — including LLM-based logic where useful
• Build and operate the customer-facing recommendation engine that turns food waste data into actionable recommendations — purchasing suggestions, anomaly explanations, operational nudges — including LLM-based logic where useful
• Train, evaluate, and iterate on models for recommendation in food waste and usage data, improving accuracy over time
• Design the features and signals — from CV/IoT data and other sources — that feed the recommendation and detection models
• Define and track the metrics that measure whether recommendations are actually useful to customers, not just whether the pipeline ran successfully
• Bring CI/CD and experimentation discipline to model and recommendation-logic changes — automated testing, staged rollout, A/B testing or holdouts, and rollback paths
• Partner with the Data Platform team to define what data and signals you need, and with other engineering teams on the inputs their systems produce
• Continuously monitor recommendation and model performance in production, and drive the fix when it degrades

What We're Looking For

• Have designed, built, or operated a recommendation system in production — one that combines multiple data sources into a single customer-facing output — not just contributed data to someone else's model
• Experience training, evaluating, and iterating on models for anomaly detection, pattern recognition, or a similar applied ML problem in production
• Experience building recommendation or personalization logic using LLMs (prompt-based scoring, retrieval-augmented generation, agent-based reasoning) in a live product, not just a prototype
• Comfortable working with production data (Python, SQL) to source and prepare inputs for your models, even if you're not the one building the underlying data platform
• Have brought CI/CD and experimentation discipline to model or product-logic changes (automated testing, staged rollout, rollback, A/B testing), with a track record of measuring whether a change actually improved outcomes
• 5 years of experience in applied ML, recommendation systems, or a closely related field
• A bias toward action

Nice to Have

• Experience with anomaly/fraud detection, forecasting, or similar pattern-detection ML problems
• Exposure to computer vision or IoT sensor data as a model input
• Familiarity with feature stores or ML feature pipelines
• Experience with Hex, Mixpanel, Tableau, or similar BI/analytics tools

The estimated base salary range for this position is $210k to $240k, which does not include the value of benefits or a potential equity grant. A wide range of factors are considered in making compensation decisions, including but not limited to skill sets, market conditions, experience and training, licensure and certifications, and business and organizational needs.