AI Engineer · Software Engineer

AI Systems × Software Engineering

Anchi Li — I build and evaluate LLM agents, code intelligence systems, and production-ready AI applications.

90%+
Human–AI agreement
100K+
Repositories analyzed
+40%
Retrieval recall
−60%
First-token latency

Selected Systems

SELECTED / 03
01

RAGenius

A production-ready knowledge system with a six-stage retrieval pipeline, concurrent execution, model caching, SSE streaming, and containerized deployment.

Python · React · LangChain · Docker

+40%
retrieval recall
Source & case study ↗
02

RepoHealth

A repository health and lifespan analysis system built from more than 100,000 GitHub repositories, combining survival analysis and deep learning.

PyTorch · Transformers · Survival Analysis

R² 0.95+
best test score
View source ↗
03

UX-Ray

A deployed web UI inspection tool that uses multimodal LLM analysis to identify usability issues and turn them into actionable recommendations.

Next.js · TypeScript · Gemini API

LIVE
interactive demo
Open demo ↗
View All Projects →

Engineering Experience

2024 — 2026
ByteDance Seed
2026.03 — 2026.06
Algorithm Engineering Intern · Code Evaluation

Built distributed pipelines for coding-agent evaluation, repository understanding, and frontend Agent-as-a-Judge; a 2K+ benchmark achieved 90%+ human–AI scoring agreement.

REALISE Lab · Concordia University
2025.07 — 2025.10
Mitacs Globalink Research Intern

Designed a RAG and multi-agent workflow for low-resource library code generation, reducing errors by up to 25%.

ISCAS · HCI & IIP Lab
2024.06 — 2025.07
Research Assistant

Engineered multi-turn conversation and memory for a low-code LLM agent platform, and improved Llama 3–8B instruction-following accuracy by 10% through LoRA fine-tuning.

View Experience →
Let's Connect

I'm interested in opportunities across AI engineering, software engineering, and applied LLM systems.

Get in Touch →