About the role
Responsibilities:
- End-to-End Ownership: The role expects candidates to take ambiguous business/clinical questions, scope them into data problems, build models, and deliver measurable outcomes.
- Core Work: Predictive modeling in supply chain forecasting, claims/denial analytics, and population health indicators.
- Advanced ML and AI: Strong emphasis on NLP, LLMs, and RAG systems, including text-to-SQL, retrieval over schemas, and grounding outputs to trusted data.
- Production Focus: Models must be production-ized pipelines, with retraining, monitoring, and drift detection, not just research prototypes.
- Collaboration: Partnering with product, domain, and government stakeholders; clear communication of findings and limitations.
- Leadership: Setting analytical standards, mentoring junior scientists, and contributing to architecture/tooling decisions.
Requirements:
- Experience: 5+ years in data science with shipped models.
- Tech Stack: Python (Pandas, NumPy, scikit-learn, and deep learning frameworks), SQL, and columnar DBs (ClickHouse, BigQuery, and Snowflake).
- Specialized Skills: NLP, LLMs, RAG, text-to-SQL, MLOps.
- Nice-to-Have: Health data (FHIR, claims, logistics), time-series forecasting, causal inference, LangChain/LlamaIndex, dbt, and lakehouse architectures.