Xuming Huang

I'm an innovative CS undergrad at University of Wisconsin - Madison, exploring the cutting edge where AI meets systems, passionate about transforming bold ideas into rigorous research. My research interest sits at the intersection of Artificial Intelligence, Systems, and Security.

Currently advised by Professor Remzi Arpaci-Dusseau and Vinay Banakar at The ADvanced Systems Laboratory, I'm developing LinuxGuard - using LLMs to automatically convert kernel bug fixes into static analyzers. Our system learns from commit history to generate clang-tidy checkers that detect similar vulnerabilities across different kernel versions, essentially turning every patched bug into a preventive tool.

I am also a remote research intern at WukLab with Professor Yiying Zhang (UC San Diego), where I work on two threads that meet at the LLM-systems boundary: (i) LLM-driven low-level code optimization, building agents that propose, verify, and benchmark compiler-grade transformations on real kernels, and (ii) system optimization for edge-device inference, profiling Apple M-series memory budgets, KV-cache behavior, and prefix-cache compaction in agentic frameworks (OpenClaw on ollama) to understand what really governs latency on a 24 GB laptop.

Last summer at Stanford (CS107: A+, CS161: A), I discovered critical access control misconfigurations in Stanford's AFS through reverse engineering exposed solution binaries. This hands-on experience, combined with my current research bridging compilers and ML, has prepared me to contribute to projects in systems security, program analysis, or ML for systems optimization.

[show email] / CV / Google Scholar / GitHub / LinkedIn

profile photo

Home  |  Research  |  Projects  |  Blog  |  Courses  |  Ask Me

Research

My research focuses on the intersection of systems security, program analysis, and machine learning. I'm particularly interested in automated vulnerability detection, static analysis generation, and applying ML techniques to enhance system reliability and security. Representative projects are highlighted.

View All Works →
Selected Projects

I create interactive visualizations and educational tools for computer science concepts. See All Projects →

Memory Hierarchy

Memory Hierarchy & Cache Indexing

VIVT, PIPT, VIPT — TLB, L1/L2/L3, DRAM, disk — one virtual address traced end-to-end
OpenClaw Workflow

OpenClaw Edge Workflow

Inside an agentic LLM turn — prompt cache, KV cache, and the 35× compaction spike
Heap Allocator

Heap Allocator Visualization

CS107 memory allocation strategies with coalescing and fragmentation analysis
Teaching & Tutorials

Deep Learning Architecture Tutorials:
• Understanding Transformers - Comprehensive tutorial on Transformer architecture with implementations
• GPT Implementation Guide - Step-by-step implementation and explanation of GPT models
• Algorithms Visualizations - Interactive visualizations and implementations of fundamental algorithms

Teaching Assistant Aspirations:
Planning to TA for COMP SCI 252: Introduction to Computer Engineering (Future semesters)

Apps
TenantSOS TenantSOS - Legal Information iOS App
Swift 5.9, SwiftUI, Firebase, Core Location

iOS application providing location-based legal information for tenants across all 50 U.S. states. Features automatic GPS-based state law detection, comprehensive legal database covering tenant rights, traffic laws, employment regulations, and consumer protections. Includes 10+ legal document templates, smart notifications for law changes, and offline law access. Implements freemium model with pro subscription for unlimited features.

[GitHub]
Service & Activities

Leadership: NFL Flag Football Team Captain - USST "Earthmoving Vehicles" (2023 National Champions)
Community: Peer tutor for CS252 Discussion Sections at UW-Madison (coming soon...)
Open Source: Contributing to machine learning projects and educational materials


Feel free to clone my templete Xuming Huang