Projects

AI has bottlenecks. I build the projects that solve them — from giving agents real-world tools to simulating entire AI-driven economies.

01
AI + Networking

AI can’t see packets

On networking projects, my AI agents could write code about packets but couldn’t see them. So I built MCP-Wireshark: now any LLM can capture live traffic, parse .pcap files, and follow streams from inside Claude.

MCP-Wireshark

39
39 GitHub Stars
PythonMCPWiresharktsharkClaude Code
How a request flows

From plain-English question → live packet analysis

prompttool calltsharkparseJSONanswer
User
User
Claude / Anthropic logo
Claude
Model Context Protocol logo
MCP-Wireshark10 tools · Python
Wireshark logo
.pcap / live
Pandas logo (data analysis)
Analysis

A real round-trip for one tool call — purple packets carry the request out, the emerald packet carries the synthesized answer back to the user. Hover any node to inspect it.

Safe by default · path + filter validation · capture caps · no shell=True
02
AI + Energy

Enterprise knowledge is trapped in docs

Engineers at SEL have decades of relay testing knowledge — scattered across manuals, test logs, and tribal memory. I’m building a RAG system that turns all of it into instant, AI-powered answers. No more digging through 500-page PDFs.

LLM-Powered RAG for Engineers

Enhancing Engineer Productivity
PythonLangChainRAGVector DBFastAPI

AI + Energy

IMAGE / VIDEO DEMO
03
AI + Finance

AI sees what humans miss

Markets move on signals humans can’t process fast enough. I built a multi-modal fusion architecture that stacks LSTM over 60 days of price data and fuses it with FinBERT sentiment from financial news — catching patterns no single model could find alone.

ContextQuant

Multi-Modal Fusion Architecture
PythonPyTorchLSTMFinBERTStreamlit
📈

AI + Finance

IMAGE / VIDEO DEMO
04
AI + Cybersecurity

AI must earn trust before deployment

You can’t deploy AI in security if you can’t prove it works. This ensemble ML pipeline evaluates 48 features to classify phishing websites with 98% accuracy. It won Best Project at UIC Engineering Expo — because trust is earned through rigor.

SecureSense

2
Best Project @ UIC Expo
PythonScikit-LearnPandasTensorFlow
🛡️

AI + Cybersecurity

IMAGE / VIDEO DEMO
05
AI + Infrastructure

LLM infrastructure can’t keep up

When every team picks a different LLM, how do you know which one actually wins? I built a high-concurrency Go CLI that races multiple LLMs head-to-head in real time — streaming results in parallel and rendering a live terminal dashboard.

RaceLLM

Real-time LLM Benchmarking
GoConcurrencySSEBubbleTeaCLI
🏎️

AI + Infrastructure

IMAGE / VIDEO DEMO
06
AI + Economics

What happens when AI runs the economy?

Nobody knows what emergent behaviors appear when you give LLM agents money, markets, and free will. This simulation explores that question — autonomous AI agents trading, negotiating, and evolving inside a dynamic marketplace.

Project Darwin

Coming Soon
Emergent AI Behavior
PythonLLM AgentsTypeScriptDockerSimulation
🧬

AI + Economics

IMAGE / VIDEO DEMO
Coming Soon

This is just the highlight reel.

See everything on GitHub