Skip to content
View Roxin-ChaI's full-sized avatar
🐰
Focusing
🐰
Focusing

Block or report Roxin-ChaI

Block user

Prevent this user from interacting with your repositories and sending you notifications. Learn more about blocking users.

You must be logged in to block users.

Content in all repositories owned by your account will be closed.
Maximum 250 characters. Please don’t include any personal information such as legal names or email addresses. This note will only be visible to you.
Report abuse

Contact GitHub support about this user’s behavior. Learn more about reporting abuse.

Report abuse
Roxin-ChaI/README.md

Hi, I'm Yixin 👋

Software Engineer building AI Agents and production-oriented LLM applications.

📍 Xi'an · 💼 AI Agent / LLM Application Engineering

Python LangGraph LangChain FastAPI PostgreSQL Redis Docker

Agent Systems · RAG · Context Engineering · AI Infrastructure

I have several years of experience building practical software systems around simulation, machine learning, and data-intensive workflows, including Qt desktop applications, model-training tools, and reliable data platforms.

I am now focused on building AI-native products and developer tools, with particular interest in AI agents, LLM-powered workflows, tool use, and knowledge retrieval. I bring a strong engineering mindset to AI development, emphasizing reliability, clear system boundaries, rapid experimentation, and evidence-driven iteration.

🚀 Projects

  • 🐴 Code Mule — Human-supervised autonomous software engineering runtime that plans, executes, reviews, verifies, and delivers coding tasks with DeepSeek + Codex.

  • 🧰 Agent Engineering Workbench — Unified full-stack workspace integrating production-oriented AI agents, research workflows, context engineering, evaluation, and prompt tooling.

  • 🤖 Production Knowledge Research Agent — Production-oriented research agent with LangGraph, hybrid RAG, grounding, evaluation, reliability, and observability.

  • 🔎 Web Research Agent — Lightweight LangChain agent that autonomously uses web search through tool calling.

  • 📄 AI Resume Optimizer — Evidence-grounded resume optimization tool that analyzes job requirements, rewrites resumes conservatively, validates factual provenance, and exports editable DOCX/Markdown results with DeepSeek.

  • 🧠 Context Window Compressor — Framework-agnostic context engineering library for managing LLM Agent history under token constraints.

  • 🔍 AI GitHub Reviewer — Read-only AI Pull Request reviewer with bounded tool calling and deterministic output validation.

  • ⚙️ Prompt Vault API — FastAPI backend demonstrating persistence, migrations, Docker, testing, and CI.

Connect

Portfolio Email

Pinned Loading

  1. production-knowledge-research-agent production-knowledge-research-agent Public

    Production-oriented single-agent knowledge research system with DeepSeek V4 Flash, hybrid RAG, pgvector, Web Search, evidence grounding, evaluation, reliability, and real E2E validation.

    Python

  2. web-research-agent web-research-agent Public

    A lightweight single-agent web research system using DeepSeek V4 Flash, Tool Calling, ReAct orchestration, and DDGS web search.

    Python

  3. ai-resume-optimizer ai-resume-optimizer Public

    A truthfulness-aware resume optimization CLI powered by DeepSeek.

    Python

  4. ai-github-reviewer ai-github-reviewer Public

    A read-only AI agent that reviews public GitHub pull requests.

    Python

  5. context-window-compressor context-window-compressor Public

    A reusable Python library for LLM context compression, token budgeting, and context-window management.

    Python

  6. prompt-vault-agent prompt-vault-agent Public

    A DeepSeek-powered single-agent CLI that uses natural-language tool calling to search prompts through the Prompt Vault API.

    Python