LLM Trainer & AI Systems Developer
Engineering real-world AI systems across cloud and production environments.
“I don’t just use AI models I train and optimize them for real people.”
Currently working as an AI Systems Developer at a tech company.
I design, train, and optimize large language models (LLMs) and build
scalable backend architectures that allow AI to perform in real-world use cases.
I write code from: 🇹🇼 Taiwan • 🇰🇷 Korea • 🇭🇰 Hong Kong • 🇯🇵 Japan
| Area | Principle |
|---|---|
| Code Style | Clean architecture • modularity • readable code |
| Design Patterns | Factory • Observer • Dependency Injection |
| Async Systems | asyncio • event-driven pipelines • task queueing |
| Testing | Unit tests • integration tests • reproducibility checks |
| Versioning | Semantic Versioning • GitFlow workflow |
| Documentation | Docstrings • API schemas • markdown specs |
| Performance | latency profiling • caching strategies • optimization |
| Error Handling | custom exceptions • structured logging • fallback handling |
| Deployment Mindset | build → measure → optimize → deploy |
- Fine-tuning, transfer learning, custom dataset pipelines
- LoRA & QLoRA training, model quantization
- Prompt engineering & attention behavior analysis
- Instruction tuning & memory-enabled agents
- Conversational AI flow design (multi-turn logic)
- Token usage analysis & cost optimization
- FastAPI • Flask • Express.js
- REST & WebSocket APIs
- Microservice design & modular architecture
- Auth systems (JWT, OAuth2, session handling)
- Rate limiting & request lifecycle management
- Backend performance profiling & optimization
- Docker, Docker Compose, Kubernetes
- CI/CD pipelines • GitHub Actions
- Reverse proxies & Nginx routing
- Redis caching • async task queues
- Load testing & traffic simulation
- Server monitoring & logging architecture
- Playwright / Selenium automation
- Headless browser fingerprint spoofing
- Human-like scrolling / click simulation
- Proxy rotation & traffic distribution
- Website scraping & API reverse-engineering
- aiohttp • asyncio • threading / multiprocessing
- PyTorch • Transformers • Whisper
- SQL (MySQL / PostgreSQL) • Redis • Firebase
- FastAPI dependency injection
- Data validation with Pydantic & Marshmallow