Senior Data Scientist

About Eno Health

Eno Health is an AI-powered platform designed to be the ultimate decision-support system for healthcare providers seeking to facilitate personalised care and long-term wellness for patients. We are building a compliant, secure, and European-sovereign biomedical AI solution for personalised healthcare. The platform streamlines practitioners' workflows and patient data processing, enabling faster, more precise clinical decisions. Role Description

You will own the data-science layer of a regulated clinical AI system: the models, the evaluation methodology, and the statistical rigour behind the knowledge graph. This is not a dashboards-and-churn-models role. The problems are causal reasoning over structured clinical knowledge, mapping messy real-world lab data onto biomedical ontologies, and evaluating a fine-tuned biomedical LLM to a standard that survives a medical-device audit. You will work with clinical, engineering and product colleagues to turn clinical knowledge into computable, testable artefacts. Expect a mix of modelling, ontology work, evaluation design, and writing, as every model decision at ENO needs to be documented, reviewed by a named human, and reproducible. Clinical sign-off gates what enters the knowledge graph; your job is to provide clinicians the statistical evidence to sign off. This is a full-time hybrid role based in the Brussels Metropolitan Area, with some flexibility for remote work. Qualifications

Advanced Python and PyTorch; experience taking models from experiment to production. Our inference stack is self-hosted (vLLM); familiarity with parameter-efficient fine-tuning (LoRA/QLoRA) of open medical or biomedical LLMs is a strong advantage. Grounding in probabilistic graphical models; Bayesian networks, causal inference (structural causal models, do-calculus, or counterfactual reasoning). Our core asset is a causal clinical knowledge graph; this is the reasoning substrate you will work on daily. Experience with property graph databases (Neo4j preferred): graph data modelling, graph algorithms (PageRank, random walks), and retrieval over structured knowledge (graph-RAG architectures). Working knowledge of biomedical ontologies and terminologies (eg. SNOMED CT, LOINC/UCUM, Mondo, HPO, or equivalents) and the practical realities of mapping messy clinical and lab data onto them. Applied statistics for model evaluation: experimental design, hypothesis testing, calibration, and error analysis and the discipline to document it. Model validation at ENO feeds a medical-device technical file (EU MDR, IEC 62304); reproducibility and named-reviewer sign-off are requirements of the job, not aspirations. Experience with large, sensitive datasets in healthcare or a similarly regulated domain; fluency in GDPR Article 9 constraints, pseudonymisation, and data-minimisation trade-offs. Master's or PhD in a quantitative field, or equivalent practical experience. Federated or distributed learning: training and evaluating models across data silos that cannot be centralised. Our sovereign architecture keeps patient data inside per-country cells; learning across cells without moving personal health data is where this platform is heading. Privacy-enhancing technologies beyond access control: differential privacy, secure aggregation, and conceptual command of homomorphic encryption (enough to reason about what is feasible, at what cost, and when it is the wrong tool). Streaming data experience (Kafka/Flink) for clinical, lab, and wearable ingestion pipelines. FHIR R4 and clinical interoperability standards. Understanding of functional or systems medicine, or P4 (predictive, preventive, personalised, participatory) medicine frameworks.

What We Offer

A foundational role in a clinical AI platform at the stage where architecture decisions are still being made and yours to influence. Competitive salary and an equity package. Hybrid working from our Brussels base, with flexibility for remote work. Direct collaboration with the founding team and clinical leadership.