A transcoding college student from Mainland China, I am rubbish, but I am moving forward.
Undergraduate @ UESTC · Biomedical Engineering (non-CS track, but clearly took a detour)
| Period | Role | Company |
|---|---|---|
| 🏭 05/25 – 09/25 | Digital Industries Software Engineer Intern | Siemens DISW |
| 🗪 12/25 – 04/26 | WeChat Cross-Platform Development Intern | Tencent WXG |
| 🤖 04/26 – Present | WeChat AI Agent Development Intern | Tencent WXG |
- AI Agent - Active
· 🤖 Agent Pipeline Engineering — building Function Call toolchains and agent dispatch strategies inside WeChat · 🎨 Real-time UI Rendering — Markdown rendering engine and chat UI components for AI assistant interfaces · 🔧 MCP Tooling — learning Model Context Protocol, tool registration, and agentic workflow patterns · ⚡ Agentic System Design — model scheduling, context management, reactive UI via MCP-driven state
C++17 AI Application Service Platform — self-built HTTP framework, production-grade pipelines
-
Self-authored HTTP framework — Reactor model (muduo), state-machine parser, middleware chain, dual-path routing
-
Standard MCP Server — JSON-RPC 2.0
tools/list+tools/call, directly connectable to Claude Desktop / Cursor -
Multi-strategy LLM routing — Factory + Strategy pattern, hot-swap between Qwen / Doubao / RAG / MCP
-
SSE streaming output — curl callback token-by-token push, zero-wait frontend rendering
-
RAG pipeline — Alibaba Bailian knowledge base integration + retrieval-augmented generation
-
Async architecture — AI calls offloaded to 8-thread pool, IO threads non-blocking;
shared_mutexLRU (500 sessions max) -
ONNX Runtime inference — MobileNetV2 image recognition with OpenCV preprocessing
-
RabbitMQ + MySQL — async decoupled message persistence
⚡ RainHTTP
Lightweight C++ HTTP server framework built from scratch
- Reactor event-driven model with epoll multiplexing
- Stateful HTTP/1.1 request parser + routing engine
- Middleware pipeline, session management, SSL/TLS support
Distributed KV storage with Raft consensus, implemented in C++
- Full Raft leader election + log replication + persistence
- Linearizable reads and fault-tolerant writes across cluster nodes
- Built to understand the engineering gap between theory and production consensus
📚 AI_Learning (Active)
Notes & practice projects from learning the AI engineering stack
- Covers: Agent architecture, LLM API integration, RAG pipelines, MCP tooling
- Ongoing — updated alongside internship at WeChat AI Agent team
📊 Engineering Snapshot
📫 Contact Me
- 📧 rainrain45032@gmail.com
- 🐙 github.com/Rain0832
- 💬 WeChat Official Account: 云Coding
"Build things that scale. Understand what you ship. Make AI actually work."