AI Python Engineer - Langchain LLMs - Clinical Data

ABOUT CITEMED CiteMed is a globally distributed team on a mission to improve clinical regulatory workflows. We move fast, we respect each other, and we trust people to own their work. If you thrive with autonomy and care about the impact of what you build, you’ll fit right in. We’re a small but growing distributed team — scrappy, fast-moving, and genuinely collaborative. There’s no bureaucracy to navigate, no waiting for permission — just smart people moving quickly toward something that matters. For the right person, this role has real room to grow. ABOUT THIS ROLE We have a production AI system built on LangChain and LangGraph that extracts structured clinical data from PDF research articles. It’s a core part of how Evidence Cloud — our regulatory compliance platform — works. We need a developer who can own this system, improve it, and scale it as the product grows. This is not a ticket-taker role. We want someone who understands the architecture, cares about how the code is written, and brings ideas to the table. If you’ve shipped real AI systems and can prove it, we want to talk to you. WHAT YOU’LL ACTUALLY WORK ON • Own and maintain our LangChain/LangGraph extraction workflows in production • Process complex PDFs — dense tables, figures, multi-column layouts, and clinical research formats that break standard parsers • Extract and structure clinical and medical data with high accuracy — the output feeds directly into regulatory submissions • Architect systems that are clean, maintainable, and built to scale as the codebase grows • Identify bottlenecks, failure points, and improvement opportunities — and bring solutions, not just observations • Collaborate with our product and dev team to integrate extraction workflows into the broader Evidence Cloud platform • Document your work clearly so the system is understandable and handoff-ready YOU’RE PROBABLY A GOOD FIT IF… • You have strong Python engineering skills and you care about clean architecture — messy code that works isn’t good enough for you • You have hands-on, production experience with LangChain and/or LangGraph — not just tutorials • You’ve dealt with the hard parts of PDF processing: irregular layouts, embedded tables, scanned documents, inconsistent formatting • You’ve shipped real AI systems and have the GitHub or demo to prove it • You work autonomously, communicate proactively, and don’t need to be chased for updates • You think about systems holistically — not just the feature in front of you, but how it fits into the broader architecture • Bonus: you have experience with clinical, medical, or regulatory data — you understand what a clinical evaluation report is and why accuracy matters HOW WE’LL MEASURE SUCCESS • Code quality: Clean, maintainable architecture that scales as the codebase grows • Extraction accuracy: High-accuracy structured data output from complex clinical PDFs • System reliability: Production system stable, well-documented, and handoff-ready • Proactive improvement: Identifies and flags issues, bottlenecks, and opportunities before being asked • Communication: Weekly updates via Slack — what shipped, what’s in progress, what’s blocked TO APPLY • Include the following with your application — applications without these will not be reviewed: • GitHub profile or relevant repositories showing production-quality Python and LLM work • At least one live demo or project walkthrough showcasing PDF processing, LLM pipelines, or similar work