Toronto, CA

Marwan
Abouzeid

Offensive Security × AI · Builder

I'm an early-career developer working across machine learning and offensive security, focused on where the two meet: AI red teaming.

→ projects → writing → github → contact
whoami.sh

marwan@portfolio:~$ whoami --verbose

focusai security · offensive · ml
langsPython · TypeScript · Bash
certworking on OSCP+
basedToronto & Abu Dhabi
status learning in public
1st Runner-Up NASA Space Apps Toronto 2025
7 open-source projects
3 languages: EN · AR · ES
OSCP+ July 2026
Selected Work

Projects

All projects →
ML × SecurityOpen source

wasel

Production-grade, open-source Arabic AI answer engine with multi-provider LLM routing, tiered auth, and Redis-backed rate limiting.

Next.jsTypeScriptVercel AI SDKSupabase
details → repo ↗ demo ↗
SecurityLab-only research

AeroFuzz

A contained Modbus TCP fuzzing lab for ICS/OT research, Scapy-built mutations against a simulated PLC, with a crash-detecting watchdog and reproducible failure capture.

PythonScapyDockerModbus TCP
details → repo ↗
ML & SystemsMy strongest piece

tensorgrad

A tiny reverse-mode autodiff engine, full training pipeline in under 600 lines, with four optimizers, GPU support, and a pytest suite cross-checked against PyTorch.

PythonNumPyCuPyPyTorch
details → repo ↗
/ Now

What I'm doing
right now.

A live snapshot of what I'm working on, updated regularly. Last updated

  • Preparing for the OSCP+ certification

    Hands-on offensive security, targeting July 2026.

  • Building open-source projects

    Machine learning systems and security tooling, in public.

  • Exploring AI red teaming

    Where machine learning meets offensive security.

Latest Write-ups

Notes from
the work.

All write-ups →
Let's talk

Open to internships
and collaborations.

Interesting problems in machine learning and security, or a team I can learn from. My inbox is open.

mka@ocadu.ca ↗ More about me