First seen by Alion on Nov 16, 2026.
We are looking for a hands-on Data Science Manager to lead a small, high-impact team of Data Scientists while continuing to contribute technically. This is not a coordination-only management role. You will help shape modeling approaches, review code and experiments, unblock productionization, and improve the operating cadence of the team.
The role sits at the intersection of machine learning, healthcare data, and programmatic advertising. You will oversee work spanning attribution and lift measurement, probabilistic identity, audience quality, benchmarking and evaluation, forecasting, engagement/CTR modeling, and other applied ML problems as the portfolio evolves.
You will also partner closely with Product, AI Engineering, Data/Platform Engineering, and external data partners as Doceree builds a more unified foundation for clinical-intent signals, HCP identity, anonymized patient-level data, claims-derived outcomes, and activation identifiers.
Key Responsibilities:
- Extract entities and attributes from heterogeneous sources - warehouse tables, registries and feeds, and documents such as literature and trial registries that must be parsed before linking.
- Design and implement probabilistic entity linking and deduplication across multiple large, inconsistent healthcare datasets.
- Own the linking pipeline end to end - pairwise and graph-based scoring, calibrated match probabilities, resolution into stable entities - working alongside existing data scientists to maintain and improve the algorithms already running in production.
- Define how link quality is measured: precision and recall, calibration, error analysis by source and segment, and confidence thresholds per downstream use.
- Link entities and attributes to controlled medical vocabularies, so specialties, conditions and products mean the same thing across sources.
- Use graph representations and graph ML as core linking tools - clustering over candidate-link graphs, centrality, community detection, link prediction, and embeddings for candidate retrieval.
- Build and maintain the data and graph estate - warehouse transformation, refresh, monitoring - and keep raising the performance of existing algorithms as sources, volumes and downstream needs change.
Qualifications Requirement: Experience, Skills & Education:
- Experience: 4+ years shipping production ML or data-science work, including hands-on entity linking, entity resolution or record linkage at scale.
- Languages: Python (production-grade, tested, type-hinted), strong SQL.
- Software Engineering: FastAPI, Docker, pytest, Git and CI/CD - code others can run, test and deploy without you in the room.
- Probabilistic Methods: match scoring and calibration, threshold selection, and evaluation where ground truth is scarce.
- Graphs: comfort treating linking itself as a graph problem - graph algorithms (connected components, centrality, community detection) and graph ML (embeddings, GNNs) - and a view on what it takes to run graphs at hundreds of millions of edges rather than in memory.
- Data & ML Engineering: warehouse-centric pipelines, ETL over large public and licensed datasets, extraction from semi-structured and document sources, scalable blocking, vector search, incremental refresh and data-quality monitoring.
- Stack: Snowflake or another distributed SQL warehouse (BigQuery, Databricks, Redshift); PyTorch Geometric or DGL; igraph/NetworkX; Splink; and a graph store you have run at scale, e.g. Amazon Neptune, Neo4j/Cypher, or an RDF/SPARQL store.
Preferred Candidate:
- Nice to Have: medical ontologies and controlled vocabularies (MeSH, UMLS, SNOMED, RxNorm); prescribing or claims data; healthcare sources such as NPPES/NPI, PubMed or ClinicalTrials.gov; biomedical NLP; AdTech experience and deduplicated reach measurement; regulated-environment work with PII constraints.
Skills
Entity Linking, Entity Resolution, Python, Record Linkage, SQL

