I work on AI-driven trading, system programming, and workflow automation. I love turning tedious processes into one-click tools, while ensuring performance and maintainability.
AI-powered crypto trading agent: Automated strategy execution, signal interpretation, and trading workflows using local LLMs (Ollama / vLLM).
Lightweight custom shell: Supports process control, piping, and system call implementation with a focus on architectural design.
Automated strategy execution, data processing, risk control, and engineering LLM agents (Ollama / vLLM).
Process / IPC / piping / syscalls, engineered for performance and stability.
Turning repetitive tasks into tools: reducing manual clicks while improving speed and consistency.
Deployed a localized RAG knowledge base on NVIDIA A800 using DeepSeek & vLLM. Engineered a production-ready Docker stack for inference and document parsing.
Conducted stress testing (1-128 concurrent requests) validating >96% accuracy and optimized runtime parameters for robust performance.
Designed a privacy-focused RAG chatbot integrating OpenAI-compatible endpoints with local vector stores. Optimized prompt tuning and knowledge chunking for high-fidelity responses.
Implemented a simplified Unix-like shell in C featuring command parsing, pipes, I/O redirection, and signal handling. Focused on multi-process orchestration and Linux concurrency patterns.
Open to collaboration, system design discussions, or engineering exchanges. Feel free to contact me directly.