Full-Stack Python/JavaScript Engineer - Automation Builder - MLOps
Summary
Build and deploy LLM-powered automation agents for pharma workflows using Python/JavaScript, PostgreSQL, vector DBs, and n8n on an NVIDIA DGX Spark AI sandbox.
I'm working with a fast-moving specialty pharma group operating across 40+ markets, with deep expertise in rare diseases including haematology-oncology. They partner with manufacturers to commercialize life-saving medicines quickly and efficiently, from regulatory approval through to patient access.
Reporting to the Head of AI & Digitalisation, you will be Employee #1 in the AI Hub. Your first mission: build our on-prem AI sandbox on an NVIDIA DGX Spark unit (128 GB RAM, 4 TB SSD), then ship 5-7 production-ready AI automations in 2026. This is a greenfield role—no legacy systems, no bureaucracy, just you, the DGX box, LLM APIs, and a mandate to prototype fast and fail forward.
You will build the core stack (n8n, PostgreSQL, vector DB, Langfuse, Prometheus/Grafana, Ollama) and create LLM-powered agents for regulatory submissions, HCP targeting, pharmacovigilance, and internal ops—automations that will 3X our capacity without adding headcount.
Requirements
- LLM integration: OpenAI/Anthropic/OpenRouter APIs, prompt engineering, function calling, streaming
- Database & storage: PostgreSQL, vector DBs (Qdrant/Pinecone/Weaviate), Redis/caching patterns
- Workflow automation: n8n, Zapier, Make.com, or similar low-code/no-code platforms
- Linux/Docker: comfortable with SSH, docker-compose, systemd, basic networking