Jiaqi Li

About

Primary engineer at Customer Insights AI who built an end-to-end Python pipeline for 340B drug pricing compliance, using ML to detect suspicious pharmaceutical claims and benefit diversion. Stands out for combining healthcare compliance domain knowledge with production reliability practices, and for turning ambiguous analyst-driven review processes into automated workflows that cut manual review by 70%.

Key Strengths

• Built end-to-end Python production integration for 340B drug pricing compliance
• Strong reliability engineering mindset for ML/data pipelines
• Designed testing strategy with unit tests, mocks, and schema validation
• Implemented resilient error handling with circuit breakers and fallback heuristics
• Applied observability practices including distributed tracing
• Translated ambiguous manual analyst workflows into automated compliance systems
• Blended ML automation with human-in-the-loop review for low-confidence cases
• Defined clear delivery metrics: sub-10-minute runtime, 90%+ unit test coverage
• Delivered measurable impact: 70% reduction in manual review
• Improved automation reliability from 30% failure rate to under 5%
• Diagnosed production-only failures involving auth tokens, rate limits, and Kubernetes environment differences
• Built and deployed production LLM-driven multi-agent analytics platform (ciATHENA) for healthcare/life sciences
• Strong focus on reliability: strict schemas, Pydantic validation, node-level output checks, fallback/repair logic
• Designed stateful, multi-step agent workflows with LangGraph (state passing, retries, session tracking)
• Performance/cost optimization via prompt optimization, metadata caching, and reducing unnecessary model calls
• End-to-end quality approach: unit + E2E testing, SQL safety checks, business-rule validation, logging for audit/debug
• Improved non-technical UX with guided NL interface, structured action/result displays, and visualizations

Experience

Education

Skills

Languages