I'm Sean, a 2027 computer science undergraduate at Minzu University of China, focused on Agent and LLM application development. I turn ideas into working systems with Python, FastAPI, structured outputs, tool calling and retrieval pipelines.
My work centers on engineering reliability rather than model research: designing controllable agent workflows, building RAG and data pipelines, adding evals, traces and guardrails, then iterating through measurable results. I also document AI-native development through Sean's Build Log.
LLM features with streaming responses, structured outputs, tool calling and context-aware prompts.
Controllable agent loops with clear state, tool boundaries, structured handoffs and failure handling.
Document extraction, retrieval and evaluation pipelines for grounded, source-traced answers.
FastAPI services, asynchronous workflows, schemas, tests and evaluation for dependable AI applications.
参与智能体侧对已有本体平台的能力接入，围绕政策规则建模任务补充 Session / Retry / Checkpoint，支持失败重试与断点恢复，并参与配置化 Skill 与动态 Tool Loading 的接入验证，使不同领域工具能够按配置加载。
An advisory PR review agent that looks beyond CI for behavioral regressions, missing tests, boundary cases and maintainability risks.
A batch CV tailoring workflow that clusters job descriptions, extracts evidence from a base résumé and generates structured, traceable variants.
A reproducible comparison of BM25, hybrid retrieval and reranking using golden queries, MRR and Recall@10, with the strongest run reaching 0.940 Recall@10.
AI Application Development Intern. Delivered enterprise AI applications end-to-end — building chatbot evaluation suites, RAG pipelines and data-integration prototypes.
B.Eng in Computer Science (GPA 3.6). Coursework across AI, databases and distributed systems, with IELTS 7.0 backing strong English for reading papers and writing technical docs.
