Wenxuan Z.

Experience

1. Engineering Manager & Tech Lead

   Meta

   Jan 2018 — Present · Menlo Park, CA, US

   Hands on Agentic AI Engineering Leadership.I support RecSys Model (PyTorch) Optimization & Generation team in AI/ML Serving Infrastructure team, focusing on enabling and optimizing ML performance from trained to serving on AI Accelerators for critical biz modelsKey Projects:a) Agentic-Based Innovation: Spearheaded the 0→1 development of an Agentic-based Inference Enablement framework, automating heterogeneous hardware workflows and accelerating enablement time from 8 months->4 weeks.b) Real-Time Serving: Revolutionized RecSys model freshness by launching real-time publishing solutions, achieving model-to-serving latency from hours->minutes for biz-critical models.c) MLOps: Led and built observability and governance for model lifecycle (testing, drift detection, rollback), improving model reliability and compliance.d) System Scaling: Scaled model optimization capabilities 20x for large-scale models while increasing publishing reliability from 90% to a sustainable 97%.e) ML Lifecycle Management: Orchestrated the end-to-end ML lifecycle, ensuring high-performance inference and efficient training pipelines across cross-functional stakeholders.f) Team Leadership: Scaled the AI/ML Inference Pillar from 6 to 20+ engineers, mentoring multiple Staff and Sr-Staff Engineers while defining the long-term technical direction for recommendation systems.

Education

• University of Chicago

  Master of Science (M.S.), Physical Sciences

• Nanjing University

  Bachelor of Science (B.S.)