Hi, my name is

Pinhao Chen.

Software engineer at NTU CSIE. I work at the intersection of AI, systems programming, and infrastructure — building things that are fast, correct, and actually ship.

See my work Get in touch

01.

About Me

Hey! I'm Pinhao, a CS student at National Taiwan University with a passion for building systems at the edge of performance and intelligence.

I work on AI-driven trading infrastructure, low-level Linux tools, and local LLM deployment. I like problems where both correctness and speed are non-negotiable.

Lately I've been exploring RAG pipelines, concurrent request handling with vLLM, and squeezing more out of C on real hardware.

Some things I work with:

Python
C / Linux
LLM / Ollama
vLLM
RAG / DeepSeek
Nginx
IPC / Syscalls
Git / Shell

02.

Projects

Chao Bi

AI Quant Trading Agent

AI-powered crypto trading agent with automated strategy execution. Integrates local LLMs via Ollama for signal generation, risk control, and real-time decisions — no data leaves the local network.

PythonLLMOllamaAPI

HSH

Custom Unix Shell

A lightweight shell written in C from scratch. Supports command parsing, process management, pipes, I/O redirection, job control, and direct POSIX system calls.

CLinuxSyscallsIPC

Localized Knowledge Database

Shanghai Supercomputer Center

Private, on-premises RAG knowledge base built for SSC. Achieved >96% answer accuracy under stress tests of up to 128 concurrent requests, with no data leaving the local network.

PythonDeepSeekRAGvLLM

FigureOut

PVC Figure Market Intelligence

Secondhand price tracking platform for PVC collectible figures in Taiwan. Aggregates 37,000+ figures with real transaction data, trend analysis, and a Chrome extension for reporting prices while browsing.

Next.jsFastAPIPostgreSQLDocker

03.

Experience

Sep 2025 – Feb 2026

AI Quant Trading Developer

Chao Bi — Personal Project

Built an AI-powered crypto trading agent with automated strategy execution using local LLMs.
Integrated Ollama for on-device signal generation and risk control without relying on cloud APIs.
Designed a modular pipeline covering market data ingestion, LLM inference, and trade execution.

Jun – Aug 2025

Testing Intern

Shanghai Supercomputer Center (SSC)

Deployed a localized RAG knowledge base using DeepSeek & vLLM on HPC infrastructure.
Stress-tested the system from 1 to 128 concurrent requests, achieving >96% answer accuracy.
Built OpenAI-compatible endpoints for seamless integration with existing tooling.

2024

Linux System Tools & Mini Shell

Academic Project — NTU CSIE

Implemented a custom Unix shell in C with pipes, I/O redirection, and job control.
Explored IPC mechanisms, process group management, and POSIX signal handling.