Healthcare AI Data Scientist (Chart Processing)

About the Role

We are looking for a Python developer or data scientist who enjoys working with AI and wants to build a career in US healthcare. You will take over and improve our medical chart processing system. It reads patient charts (PDFs), extracts the text, and uses LLMs to find diagnoses and their ICD-10 codes for coders to review. You do not need deep healthcare experience; we will teach you the coding rules. You do need solid Python and real curiosity about AI.

Key Responsibilities

• Maintain and improve the Python code that processes medical charts: PDF text extraction and OCR, splitting charts by visit, and extracting diagnoses.
• Write, test, and improve LLM prompts (OpenAI, Azure OpenAI, Claude) that return structured JSON.
• Measure accuracy against coder-reviewed samples and fix the errors you find.
• Validate extracted codes against ICD-10 and other standard code sets.
• Store and query data in MongoDB; write data checks and simple reports.
• Deploy and debug on Azure (Functions, Blob Storage, queues) using Docker and Git.
• Document the system and work closely with coders and engineers in a small team.

Requirements & Tech Stack

• Good Python: pandas, files and JSON, writing and debugging scripts, Git.
• Basic ML / NLP and GenAI knowledge: you have called an LLM API or built an NLP or ML project. College projects count.
• Basic database skills: SQL or MongoDB.
• Basic healthcare awareness: what ICD-10 codes are and comfort with medical terms, or a clear willingness to learn them fast.
• Good written English.

Nice to Have

• Exposure to US healthcare data: ICD-10, CPT/HCPCS, claims, or HL7/FHIR.
• RAG, embeddings, LangChain, or similar frameworks.
• OCR and PDF tools (Tesseract, PyMuPDF).
• Cloud (Azure preferred), Docker, FastAPI.
• Medical coding coursework or certification (CPC).
• Some overlap with US time zones.