Backend Engineer | Agent Systems | AI Application Backends | Developer Workflows
- Building backend services and Agent systems with Python, FastAPI, and WebSocket
- Interested in reliable Agent workflows, Multi-Agent coordination, RAG, and evaluation loops
- Experienced with task routing, tool calling, structured outputs, observability, and Bad Case regression
- Comfortable with Linux, Docker, PostgreSQL, Redis, Elasticsearch, Milvus, and Neo4j
- M.S. student in Computer Science at Yanshan University, graduating in 2027
Python Go FastAPI WebSocket SQL
RAG LangGraph PyTorch Transformers Pydantic
PostgreSQL Redis Elasticsearch Docker Linux OpenTelemetry GitHub Actions
πΊ AI Werewolf
A multi-agent Werewolf game and evaluation system built around a Python backend. It combines a rules engine, role-based Cognitive Agents, strict information isolation, WebSocket interaction, post-game decision scoring, structured reviews, and strategy knowledge feedback.
Backend focus: FastAPI, PostgreSQL, SQLAlchemy, WebSocket, state-machine driven game flow, GameState β PlayerView projections, decision audit trails, Track B evaluation, and Track C strategy retrieval.
π§° OpenClaw Auto Skills
A reusable automation and workflow utility repository for developer tooling. It explores how repeated operations can be organized into portable skills and execution workflows for AI-assisted development.
Backend focus: Python automation, CLI workflows, reusable procedures, and developer productivity infrastructure.
π OpenLLMetry
A fork of the OpenLLMetry project for OpenTelemetry-based observability in GenAI and LLM applications, including instrumentation for model and vector-database calls.
Agent focus: tracing prompts, model calls, tools, retrieval steps, latency, and failures across AI application backends.
Backend engineering Β· Agent systems Β· Multi-Agent orchestration Β· RAG / GraphRAG Β· LLM observability Β· Evaluation and feedback loops Β· Developer tooling Β· Reproducible workflows