Projects

Open Source · No GPU Required · Python + PyTorch

Building an LLM from Zero

Working Book  ·  So Simple You Can Teach It to Your Kids

A step-by-step tutorial for building a GPT-style language model entirely from scratch using Python and PyTorch. 17 chapters across 6 parts. No GPU or cloud account needed — runs on any laptop in about 20-30 minutes.

Learn More Read Online View Source Code
AI Security · LLM Vulnerabilities · Real-Time Monitoring · Privacy · Open Access

AI Sec Watch

aisecwatch.com  ·  Launched 2025

AI Sec Watch is a tool I built to help keep track of what's happening in AI security. It aggregates incidents, vulnerabilities, and developments from across the AI ecosystem in one place.

It covers LLM exploits, agentic AI risks, jailbreaks, privacy breaches, policy changes, and more. Items are categorized and searchable so you can focus on what's relevant to you.

What it tracks

LLM vulnerabilities and exploits
Agentic AI attack vectors
AI safety research and incidents
Privacy breaches and data risks
Generative AI jailbreaks
Government and procurement policy
Model releases and capabilities
Daily briefings and newsletter
Visit AI Sec Watch