AI Engineer

Apollo gateway; follow cross-region inference and model-routing conventions defined by the central platform team - Data access: implement purpose-bound data scoping (objects, attributes, row-level) per user function; align with the data-platform team on the chosen mechanism (Snowflake RBAC, row-level security, MCP, or a combination) - Integrate the agent with central semantic assets owned by the Data Foundation team: business glossary, KPI definitions, semantic views, ontology references. - Contribute use-case-specific semantic enrichments back into the central layer. - Design for re-hookability as the OMA warehouse migrates into the Medicine Data Foundation. - Design and operate a structured evaluation harness: SQL correctness, answer faithfulness, regression on a golden-question set. - Maintain a verified-query repository for OMA clinical use cases (CDO, QM, RA domains). - Apply statistical rigor: confidence intervals, regression gates, drift detection. - Run adversarial

red-team testing in close loop with clinical analysts. - Observability for agent traces using one chosen tool (LangSmith or equivalent): plan generation, tool calls, intermediate outputs. - Monitor latency, token cost, error rates; collaborate with platform and FinOps on cost attribution and quota policy. - IT EDP Medicine Data & Analytics What we are looking for - Strong experience with Python at production level. - Advanced SQL skills, including analytical queries and optimization - Prompt engineering, function calling / tool use, ReAct or plan-and-execute patterns. - Demonstrable experience building text-to-SQL or conversational-analytics systems. - Working knowledge of Snowflake (data modeling, SQL, basic Cortex concepts). - Agent-orchestration capability, either: LangGraph (for custom-build track) OR a vendor agent platform such as Snowflake Intelligence, Cortex Analyst, or comparable. - Working experience with structured evaluation frameworks (Ragas OR DeepEval) and/or observability tools (LangSmith, Weights & Biases, Arize, or equivalent). - Working in regulated, governance-first environments: RBAC, data classification, audit trails, responsible-AI guardrails. - Advanced English Level. Nice To Have - Snowflake Cortex deep experience: Cortex Analyst, Semantic Views, Cortex Search. - Apollo / internal-LLM-gateway integration experience, or willingness to ramp quickly. - Clinical-operations domain awareness (a paired OMA clinical analyst is part of the working setup). - Knowledge graph / ontology experience (relevant for Phase 2 when EKG integration becomes active). - Infrastructure-as-code for Snowflake (Terraform, Schemachange, SnowCLI). - Containerization basics (Docker, Kubernetes) for custom agent services on the build track. - Familiarity with FAIR data principles, CDISC, or DDF What We Offer - Permanent contract through Hays. - Flexible remote working model, with one day per month on-site at the office in Sant Cugat del Vallès, Barcelona. - Competitive salary package. - Work on a high-impact internal platform used by global teams - Join a stable, innovation-driven pharmaceutical environment with a long-term mindset - Opportunity to contribute to automation and future AI initiatives in a regulated, meaningful industry If you're passionate about building impactful solutions and want to be part of a collaborative, forward-thinking team, we'd love to hear from you.