Building reliable AI agents, practical LLM applications, and verifiable workflows.
🎓 Wuhan University of Technology ('28) · 📍 Wuhan, China · 🎯 Seeking AI Agent / LLM App Internships
- 🤖 Focus: Building reliable AI agents, tool-using workflows, RAG systems, and developer productivity tools.
- ⚙️ Engineering: Plan-and-Execute orchestration, structured outputs (
Pydantic), failure handling, and benchmark evaluations. - 🌟 Open Source: Upstream contributor to n8n, DeepTutor, and developer tooling ecosystems.
Python (asyncio / Pydantic / pytest) · Agent Workflows & Tool Calling · Hybrid RAG & RRF · CI/CD & Mocking
| Project | Description | Tech Stack |
|---|---|---|
| CountyResearchAI | LLM Research Agent: Plan-and-Execute runtime, dynamic question-tree planning, multi-search provider routing, and 260+ regression tests. | Python Agent Runtime pytest |
| courselm | Document Automation Pipeline: Slides parsing → LLM distillation → Pydantic validation, featuring per-stage caching, resume-on-break, and Windows CI. | Python Pydantic CI/CD |
Other Projects: MicroGlow-Translate (ESP32 TinyML, Provincial Grant) · esp-screenpilot (ESP32-S3 HID) · student-os
- n8n-io / n8n (Workflow Automation Platform)
- Fixed error payload loss and
TypeErrorunder "Continue On Fail" with HTTP regression tests (#40368 · Merged).
- Fixed error payload loss and
- HKUDS / DeepTutor (AI Tutoring Assistant)
- Draculabo / AntigravityManager (Model API Gateway)
- Resolved HTTP 413 payload truncation on multimodal inference via 64 MiB routing limit (#335 · Merged).
⭐️ Thanks for visiting! All code, benchmarks, and PRs are verifiable across the linked repositories.