AI Engineer

We’re building the future of healthcare automation — starting with the  post-acute, value-based care space , where both medication overload and revenue cycle complexity are driving unnecessary costs and poor outcomes. Founded by Dev Roy (2x exited founder), our stealth startup is building  multi-agent LLM systems  that operate with  transparency, accuracy, and traceability  — tackling two of the hardest problems in the healthcare stack :   medication optimization  and  revenue cycle management (RCM) .

We use  LangGraph  and a tightly structured  MCP (Model-Context-Protocol)  architecture — where agents don’t just generate text, but operate over  stateful contexts , issue  tool calls , and follow  explicit protocol-guided reasoning loops . The agents we build don’t guess — they explain, document, and prove.

Responsibilities

Architect and implement  multi-agent systems  using LangGraph, leveraging an  MCP-inspired structure : agents with memory, contextual tool use, and protocol-defined responsibilities

Build workflows that replicate expert-level medical coding and medication review processes

Design tool interfaces (e.g. vector DB, ICD-10 knowledge graph, FHIR endpoints) and integrate them into agents’ decision-making loops

Orchestrate LLM calls with clear traceability, structured state, and logging for regulatory-grade auditability

Own experiments and iterate rapidly to improve reasoning, accuracy, and output alignment

Collaborate closely with the founder on roadmap, architecture, and product direction

Qualifications

1+ years of experience in AI / ML, NLP, or backend system design

Strong Python engineering fundamentals and practical experience with  LangGraph ,  LangChain , or similar agent frameworks

Comfortable building  MCP-style architectures  — where agents reason with structured inputs, manage context, and follow execution protocols

Experience with  LLM APIs  (OpenAI, Anthropic, Claude), prompt engineering, and managing token / latency trade-offs

Familiarity with  vector search engines  (Pinecone, Chroma, Weaviate) and  RAG-based retrieval systems

Bonus : background in  clinical NLP ,  medical coding standards  (ICD-10, CPT), or exposure to  FHIR / HL7

Startup mindset : autonomous, fast-moving, curious, and deeply pragmatic

If you want to build  human-grade AI agents  that actually matter — not chatbots, but reasoning systems that impact clinical decisions and billing integrity — we’d love to talk.