Full-Stack Engineer (React Flow + Python/FastAPI)

SageMed builds software that turns scientific data into the documentation life sciences companies depend on: scientific validity reports, clinical evaluation reports, claims substantiation, market and competitive research, and more. Our platform, SageMed Bridge, does this work as a set of modules over a shared evidence and traceability layer. The work it replaces takes specialist teams months, and the output has to survive a notified body, a regulator or a legal reviewer reading it line by line. We're looking for a senior full stack engineer to join the small team building Bridge end to end: the Next.js interface specialists work in, the services behind it, the pipelines that ingest and process evidence, and the agentic AI workflows underneath. What you'll work on: You could be building the one module one quarter and something entirely different the next. The modules may vary, the underlying problem does not. User-facing product. Next.js and TypeScript. Review, appraisal, evidence-to-claim mapping, and document output that specialists live in for hours at a time. Backend services and data pipelines. Ingesting scientific data from bibliographic databases, structured sources and PDFs, deduplication, structured extraction, versioned storage of every artifact we produce. Agentic workflows. The retrieval, appraisal, analysis and drafting agents, and the evaluation harness that tells us whether a change made them better or worse. The traceability layer. Every sentence in a generated report resolves to a real source. Every human decision is recorded with who, when and why. Every run is reproducible against a pinned model and configuration. Making it defensible. Secure and auditable enough to pass procurement and supplier qualification at large regulated manufacturers. What we're looking for: Strong TypeScript with React and Next.js, plus genuine backend depth ([our stack: e.g. Node/Python, Postgres, cloud]). You've shipped LLM-backed features that had to be correct, not just plausible, and you can describe how you measured that. Eval harnesses, regression suites, grounding checks, failure analysis. Security-by-default habits: authorisation and tenant isolation, secrets handling, dependency and supply chain hygiene, least-privilege cloud IAM, and a healthy suspicion of untrusted input, including the arbitrary PDFs we feed to agents. Comfort owning a problem from design through production with limited scaffolding. Useful, not required: GDPR in practice: data residency, DPAs, subprocessor management, records of processing. SOC 2 or ISO 27001 from the building side rather than the auditing side. A regulated background of any flavour: IVDR/MDR, ISO 13485, GxP and 21 CFR Part 11, clinical research, or similar. Life sciences domain knowledge is a real advantage. Retrieval and search at depth: hybrid retrieval, ranking, relevance evaluation. Document engineering: PDF and table extraction, structured output, deterministic document generation. Why this role is worth your time: You're close to the product and close to real users. Regulatory, clinical and commercial specialists, not a persona in a deck. Correctness is legally consequential here. A fabricated citation isn't a bad user experience, it's a finding. That constraint makes the engineering more interesting, not less. You pick the right tool for the job. There's little historical architecture to work around, and you'll have real influence over what replaces it as the team grows. Polished interfaces and solid services behind them, built by the same person. Real autonomy at an early stage, with the ownership that comes with it.