Senior Data Scientist, AI Retrieval Systems

Responsibilities

• Model biomedical knowledge for rare disease research by ingesting disease and phenotype ontologies and controlled vocabularies into PostgreSQL, reconciling identifiers, and maintaining release and refresh workflows.
• Build retrieval-augmented services that map everyday language to clinical concepts using embeddings, candidate retrieval, contextual model disambiguation, structured outputs, and validation.
• Tune keyword and vector search over large biomedical corpora, balancing recall and latency.
• Build ranking and relevance layers with domain-aware weighting and graceful degradation outside curated coverage.
• Deliver user interfaces with Next.js, React, and TypeScript, including question and confirmation flows, result presentation, and live pipeline status.
• Deploy and operate services on NIH on-premises and high-performance computing Kubernetes environments, including Helm charts, StatefulSets, secrets, ingress, GPU scheduling, and scheduled jobs.
• Build evaluation and regression systems using golden sets and retrieval metrics such as Recall@K and MRR.
• Instrument systems with request identifiers, latency, errors, and selected concepts for reviewable AI-assisted results.
• Translate researcher, clinician, and patient needs into data models, retrieval behavior, and interface design.
• Contribute to manuscripts, conference abstracts, and posters with NIH investigators.

Requirements

• Bachelor’s degree in Data Science, Computer Science, Bioinformatics, Biomedical Informatics, or a related field; equivalent professional experience may be considered.
• At least 5 years building and operating production software or data systems, including at least 2 years shipping LLM-powered applications.
• End-to-end retrieval systems experience, including indexing, query construction, and retrieval-quality measurement.
• Experience evaluating systems without a single correct answer using golden sets, offline regression suites, Recall@K, MRR, or similar metrics.
• Experience with structured output and tool or function calling using typed schemas and validation.
• Ability to own services from schema design through deployment and operation.
• Ability to obtain and maintain a Public Trust Security clearance.
• Proficiency with Python, FastAPI, Pydantic, pytest, PostgreSQL, vector and full-text search, embedding pipelines, containers, Kubernetes, Next.js, React, TypeScript, Git, and CI/CD.

Nice-to-haves

• Biomedical ontologies and controlled vocabularies, including MONDO, HPO, UMLS, MeSH, and OBO Foundry resources.
• Entity linking, concept normalization, ontology alignment, or embedding-based domain grounding.
• Helm and deployment to on-premises or HPC Kubernetes environments.
• Production serving of open-weight models with Ollama or vLLM, LiteLLM, and domain embedding models such as MedCPT.
• Rare disease, clinical genetics, or translational research experience.
• Contributions to biomedical resources, standards, or consortia such as OBO Foundry or GA4GH.
• Published or presented engineering work, open-source contributions, or prior NIH experience.

Compensation and Benefits

• Annual salary: $130,000–$150,000.
• 100% employee medical, dental, and vision coverage.
• Paid time off and holidays.
• 401(k) match up to 5%.
• Educational benefits, employee referral bonus, and flexible spending accounts.