
Hank Z.
Fullstack Developer, Microservices, RAG and LLM AI Solutions
Habilidades

Conheça meus serviços

Experiência profissional
Model Manager
Baidu • Período integral
Mar 2014 - Dec 2025 • 11 yrs 9 mos
Established and led a 10-member full-stack team using agile methods to build Model Manager, an AI model management platform covering the full model lifecycle: registration, version control, deployment, scoring, performance monitoring, and retraining. Key responsibilities: • Team leadership: Led a 10-member full-stack and QA team. • Stack modernization: Upgraded frontend from Flex/OpenUI5 to TypeScript/React and backend from Java Spring to Golang; led core modules including model scoring, performance monitoring, version control, and containerized release. • Cloud-native transformation: Migrated applications from Tomcat to Docker and Kubernetes with CI/CD pipelines, raising release frequency from annual to monthly feature releases and quarterly stable releases. • One-click deployment: Enabled one-click model deployment to local servers, cloud, and databases via Kaniko and BuildKit, publishing to Docker and GCP, supporting both real-time API scoring and offline batch prediction. • Model monitoring: Built drift detection for data and concept drift, tracking PSI, KS, and Gini metrics with an alerting mechanism triggering calibration, validation, and retraining. • DevOps & testing: Covered functional, interface, unit, integration, and regression testing; authored automated test cases, executed runs, logged and tracked defects to closure; built a JMeter + InfluxDB + Grafana performance-testing platform and containerized CI/CD testing on GitHub Actions, managing suites via Jira + TestRail. • RAG & AI insight: Built a RAG knowledge base with Python LangChain, PostgreSQL vector DB, and Azure OpenAI, plus an MCP server exposing monitoring metrics to an LLM for intelligent performance insights.