At the Institute of Process Engineering, Chinese Academy of Sciences, I designed and developed an internal domain knowledge Q&A platform covering more than 1,000 papers. It uses knowledge-graph-assisted retrieval for extraction, lithium-battery, and related domains.
I also designed a document intelligence pipeline after evaluating OCR and document-parsing approaches. I built parsing flows for complex tables and scanned or skewed pages, producing structured Markdown and JSON for downstream RAG workflows.
At the Institute of Software, Chinese Academy of Sciences, I led delivery of two cloud platform projects and contributed to two others, working directly with customers on requirements and implementation plans. I deployed and maintained Kubernetes clusters across offline environments and implemented multi-cluster management with Crossplane.
At AInnovation (Beijing) Technology Co., Ltd., I developed backend services and LLM applications, including a LangChain prompt service and a document Q&A workflow with a vector database. I also customized LangFlow for drag-and-drop agent and RAG workflows and packaged a Python SDK for reuse.
