Staff AI Engineer - Zocdoc - Remote
(2025-11 - 2026-07)
- Architected multi-agent orchestration infrastructure for Zocdoc's provider-matching and scheduling platform, embedding agentic workflows into patient-facing surfaces with planner-executor patterns coordinating search, ranking, and availability agents, reducing patient-to-appointment latency 34%
- Designed and productionized RAG pipelines over provider profiles, insurance networks, and availability data using LangGraph for agent coordination and pgvector for hybrid search, achieving 89% retrieval precision on golden-set evals
- Built automated evaluation harnesses with LLM-as-judge frameworks and regression suites covering prompt injection defense, hallucination rate, and citation grounding, establishing release-gate criteria that caught 17% of candidate models before production
- Owned multi-tenant data segmentation and HIPAA-compliant context isolation for patient-specific agent workflows, architecting secure execution boundaries across 2.7M monthly active users with zero cross-tenant leakage incidents
- Implemented SSE streaming pipelines for real-time agent responses in conversational scheduling surfaces, optimizing TTFT 420ms → 210ms and maintaining p95 latency under 380ms across peak load
- Established prompt engineering standards and structured-output validation with Pydantic schemas, reducing malformed LLM responses 19% and enabling reliable downstream workflow automation
- Mentored three mid-level engineers on agentic reasoning frameworks, tool calling patterns, and production observability practices, pairing on architecture reviews and LangSmith trace analysis
- Collaborated cross-functionally with product, data engineering, and compliance teams to define knowledge-graph requirements and continuous-monitoring signals for provider data freshness, supporting quarterly compliance audits
Applied AI Lead - United Health Group - Brentwood, TN
(2022-11 - 2025-10)
- Led end-to-end delivery of LLM-based multi-agent workflows for United Health Group's care-coordination platform, shipping retrieval-augmented claim-resolution agents that reduced manual review queues 31% across 18-month tenure
- Designed knowledge-driven applications integrating RAG over clinical guidelines, formulary data, and prior-authorization rules using LangChain and Pinecone, achieving 84% answer accuracy on member-services golden sets
- Built scalable AI inference pipelines on AWS Bedrock (Claude, GPT-4o) and SageMaker, serving ~4,100 requests/day with p95 latency under 1.2s and cost discipline that trimmed LLM spend 23% through model-tier routing
- Owned automated evaluation infrastructure with regression harnesses, golden-set versioning, and drift-detection monitors, establishing eval gates that blocked two high-hallucination-rate model updates before member-facing rollout
- Implemented multi-tenant system design patterns for secure data segmentation across payer, provider, and pharma workflows, architecting role-based access controls and audit logging for SOC 2 compliance
- Collaborated with SRE and data-engineering teams to design real-time streaming architectures (Kafka, Lambda) supporting resource-intensive AI workflows at healthcare scale, processing 1.8M events/day
Senior Software Engineer - United Health Group - Brentwood, TN
(2019-09 - 2022-10)
- Built production data pipelines and distributed backend services for United Health Group's member-engagement platform, processing insurance claims, eligibility data, and benefit lookups with 99.6% uptime across 3-year tenure
- Designed RESTful APIs and microservices in Python (FastAPI) and Go, supporting member-facing surfaces serving 6.2M active accounts with p95 API latency under 240ms
- Owned unit-testing and integration-test infrastructure, raising backend test coverage 68% → 91% and establishing CI/CD gates that reduced production incidents 28%
- Contributed to on-call rotation and incident-response runbooks, resolving data-pipeline failures and API-timeout issues with mean-time-to-mitigation under 37 minutes
- Collaborated with product and design teams on feature prioritization, translating business requirements into scalable technical designs across quarterly planning cycles
Software Engineer - IXL Learning - San Mateo, CA
(2017-07 - 2019-08)
- Developed interactive learning modules and assessment pipelines for IXL Learning's K–12 educational platform, shipping adaptive question-generation logic serving ~890K students daily
- Built backend services in Python (Django) and JavaScript (Node.js) supporting real-time progress tracking, personalized recommendations, and instructor dashboards with sub-300ms response times
- Designed data-ingestion pipelines for curriculum content and student-performance analytics, integrating third-party assessment APIs and maintaining data consistency across 14 subject areas
- Participated in code reviews and pair-programming sessions, contributing to team knowledge-sharing on asynchronous Python patterns and test-driven development practices
