Machine Learning Engineer (Foundation Models & Personalization)

Eight Sleep•San Francisco, CA

About The Position

We’re looking for a Machine Learning Engineer to build and ship consumer-facing AI systems that power personalization, coaching, and next-generation “sleep intelligence.” You’ll work across data, modeling, product, and engineering to translate research into reliable, measurable improvements for members. This role is ideal for someone who loves end-to-end ownership: from problem framing → prototyping → offline evaluation → online experimentation → production deployment → iteration.

Requirements

2+ years building ML systems in production, ideally for consumer-facing products.
Strong ML fundamentals across supervised learning, sequence/time-series modeling, and modern deep learning.
Hands-on experience with large-scale model training and evaluation (PyTorch/TensorFlow/JAX), and strong Python engineering practices.
Experience with personalization systems (ranking/recommendations, segmentation, lifecycle modeling, propensity/behavior modeling, causal/experiment-aware thinking).
Fluency with data tooling (SQL, distributed compute such as Spark/Ray, and cloud storage/compute).
Strong product sense: you can translate ambiguous goals into measurable outcomes and iterate quickly with stakeholders.

Nice To Haves

Experience applying LLMs/foundation models to product features (tool use, retrieval, structured outputs, guardrails, evals).
Experience with multimodal data (sensor signals + context) and/or health/biometrics data.
Experience with privacy-preserving approaches (on-device/federated learning, differential privacy, data minimization).
Experience designing experimentation frameworks or causal inference approaches for personalization.

Responsibilities

Build and deploy ML models that improve sleep experiences through personalization, prediction, and behavior understanding (e.g., readiness forecasting, event detection, individualized recommendations).
Apply and adapt foundation-model capabilities to real product workflows (LLM + tools/RAG, multimodal modeling, policy learning), including MCP-style integrations where helpful.
Develop user behavior models that connect longitudinal signals (sleep, environment, routines) to actionable interventions - grounded in robust experimentation and measurement.
Design evaluation strategies (offline metrics, slice-based analysis, calibration, reliability, fairness) and partner with Product to run high-quality online experiments.
Productionize models: scalable training/inference pipelines, model monitoring, drift detection, alerting, and continuous improvement loops.
Collaborate with cross-functional partners (Product, Mobile, Backend, Clinical) to scope requirements and ship high-impact features.