Xu Cheng
I am an AI technology leader with 15 years of R&D and management experience across fintech, content safety, and e-commerce. My work focuses on turning advanced AI capabilities into reliable products and scalable platforms.
LLM/VLM/RL · Agentic AI · Cloud-native Architecture · Engineering Leadership
Focus
- Agentic AI systems (planner-executor workflows, tool orchestration, production reliability)
- LLM/VLM applications (RAG pipelines, model adaptation, AI feature integration)
- Platform architecture (Kubernetes, microservices, high-concurrency distributed systems)
- Leadership (0→1 team and platform building, cross-functional execution)
Selected Impact
- Designed and delivered an agentic RAG system for compliance knowledge workflows, including document ETL pipelines.
- Led cross-functional engineering teams (up to 70 members) to deliver insurance, wealth management, and banking platform systems.
- Built a K8S cloud-native architecture that became an internal technical benchmark.
- Built a 0→1 credit card issuance and risk-control platform; achieved PCI-DSS and reduced procurement costs by millions.
- Led high-scale architecture supporting up to 60K orders/sec peak traffic.
Open Source
CLaRa-Qwen3 — RAG feature extraction model based on Qwen3-4B, designed to improve context understanding in retrieval-augmented pipelines.
Tags: Qwen3-4B · RAG · Transformers · Feature Extraction
Education
Illinois Institute of Technology — M.S., Artificial Intelligence (Vision & Control), 2022–2024
Nankai University — B.S., Software Engineering, 2004–2008
Working Style
Entrepreneurial execution · Strategic vision · Candid communication · Continuous learning