Software Developer @ Lightning Minds
May 2025 – Present- Built a LangGraph router that infers clinical role and dispatches queries across FHIR, X12, and CDISC healthcare datasets, achieving 3.8s P50 end-to-end latency.
- Deployed a serverless 311 support agent on AWS Bedrock AgentCore with OAuth2-based MCP tool integration.
- Built a RAG layer on ChromaDB (EC2-hosted) — embedding pipeline plus retrieval tuning — reaching 0.92 answer relevance on LangSmith evals.
- Developed an 8-stage e-commerce AI agent with LangGraph and FastAPI, achieving a 96% task completion rate across 150+ LangSmith eval runs.
- Instrumented the agent pipeline with OpenTelemetry on CloudWatch for distributed tracing across LLM calls, tool invocations, and retrieval.
AI Model Evaluator @ DataAnnotation
Jun 2024 – May 2025- Evaluated LLM outputs across code generation and reasoning tasks; developed systematic rubrics for identifying failure modes that informed model training improvements across multiple languages and task types.