Senior AI Engineer - Healthcare Platform

About the role

Harris is modernizing its platform and adding AI capabilities across clinical and revenue cycle workflows. It is forming a small AI engineering team in Costa Rica and is looking for a Senior AI Engineer who writes solid production code and wants to build real AI systems. You will work closely with the AI Solutions Engineer, take ownership of significant parts of the platform and help the mid-level engineers.

Key responsibilities

• Build AI agents and multi-agent workflows (tool calling, orchestration, state and memory management, retries, human-in-the-loop steps) with LangChain/LangGraph or similar.
• Build MCP servers and tools so agents can safely use CareTracker data and services.
• Develop components of the AI gateway on LiteLLM (routing across Claude and Azure OpenAI, fallbacks, budgets, cost tracking, PHI redaction) and the API gateway (authentication, rate limiting, versioning).
• Instrument agent pipelines with OpenTelemetry and Langfuse (traces, evaluations, latency, cost and quality metrics).
• Write production code in Python.NET (C#) and TypeScript, with tests, code reviews and documentation.
• Integrate with CareTracker services and healthcare data (Oracle, Postgres, HL7, FHIR, claims) while following HIPAA and PHI rules.
• Review code, pair with mid-level engineers and take the lead on features end to end.

Required profile

• 5+ years of software development experience and 5+ years of hands-on Python coding.
• 2+ years coding .NET (C#) and 2+ years building applications with LLMs (Claude, Azure OpenAI or similar).
• 2+ years building agents with LangChain, LangGraph or similar, and 2+ years with MCP, LiteLLM and observability tools.
• 1+ year with TypeScript and 1+ year in healthcare, including HIPAA and PHI handling.
• Team player with strong analytical, troubleshooting and problem-solving skills, able to work independently and speak in front of customers.

Required skills

• REST APIs, OAuth2, JWT and API security
• Git, CI/CD, containers and cloud services (Azure preferred)
• Additional: FHIR/HL7, RAG, vector databases, infrastructure as code (Bicep or Terraform)