Python Backend Engineer building production AI systems and agent workflows.
I am a Python systems engineer with 5+ years of backend experience and an MSc in Data Analytics & Artificial Intelligence. My current direction is the bridge from backend engineering into AI systems engineering: reliable APIs, typed contracts, maintainable service boundaries, agent workflows, and infrastructure that can support AI products beyond the demo stage.
I am not positioning myself as a pure ML researcher or Kaggle-style data scientist. My strongest signal is practical software engineering: clear architecture, tests, documentation, integration boundaries, and systems that remain understandable after the first version ships.
These repositories are intended to read as one portfolio story:
- Upload Control Plane shows production-oriented AI/robotics data-ingestion infrastructure: resumable multipart uploads, FastAPI service boundaries, PostgreSQL lifecycle state, MinIO/S3 adapter design, direct-to-storage upload flows, and Codex-readable PRD/task execution artifacts.
- ExcelAlchemy shows reusable Python library engineering: typed workbook contracts, Pydantic models, storage boundaries, documentation, examples, and contract tests.
- SpeechX / LingoTask applies that engineering style to a small LLM-backed FastAPI service with request/response schemas, prompt handling, Word export, and tests around deterministic behavior.
- Codex Skills captures reusable AI-assisted engineering workflows for diagnosis, TDD, architecture review, PRD writing, and durable decision traces.
- DingTalkOAuth demonstrates backend integration work around OAuth, messaging, deployment boundaries, and external platform APIs.
- Python3 Project Code Conventions documents how I think about code quality, review judgment, readability, extensibility, and robustness.
Together, these projects are meant to show a backend engineer moving toward AI systems engineering through production-minded building blocks: data ingestion control planes, contracts, tests, workflow design, maintainability, and operational awareness.
- Python backend services and libraries with stable interfaces, typed data models, and clear integration boundaries.
- AI and robotics data infrastructure where large file ingestion, storage boundaries, lifecycle state, authorization, observability, and recovery behavior need to be designed as one system.
- LLM application surfaces where prompts, schemas, provider calls, document output, errors, and tests need to fit into a maintainable service.
- Agent-assisted engineering workflows that make debugging, review, planning, and refactoring more repeatable.
- Developer-facing documentation that explains decisions, constraints, and usage paths rather than only listing commands.
I care about software that survives real constraints:
- explicit contracts at module, API, and storage boundaries
- tests that protect behavior rather than implementation accidents
- error handling and failure modes that are visible to callers
- documentation that makes architecture and tradeoffs inspectable
- maintainability over cleverness
- AI features treated as systems, not isolated prompts
- Backend: Python, FastAPI/Flask, REST/GraphQL APIs, Pydantic, SQLAlchemy, PostgreSQL, Redis
- Systems: internal SDK design, typed workflows, modular architecture, storage integration, configuration management
- AI engineering: AI/robotics data ingestion, LLM service integration, prompt workflows, document generation, agent workflows, evaluation-aware development
- Engineering practice: testing strategy, code review, refactoring, debugging, technical writing, Git-based delivery
- Native Chinese
- Fluent English
- Comfortable documenting complex systems and explaining technical decisions across teams
My goal is consistent across backend infrastructure, AI tooling, and system-level Python code:
build software that remains understandable and reliable long after it is written.