Selected work
Three things I shipped
-
CAP3D-Viz
Open-source 3D visualization for chip-capacitance data — packaged, versioned, DOI-citable.
- PyPI
- DOI
- CI/CD
- MIT
- Problem
- Capacitance solvers in IC design emit huge 3D-geometry files that are hard to inspect; most tooling stays trapped in one-off notebooks.
- Built
- A Python package with a state-machine streaming parser and interactive 3D rendering, shipped to PyPI with a full CI/CD pipeline (test, build-and-release, PyPI install verification, docs) and a registered DOI.
- Result
- Streams 10,000+ geometry blocks under 8 MB of memory; published, versioned, CI-tested, and citable via DOI.
-
FaultPilot — ArduPilot fault-injection framework
A plugin-based test harness for drone flight simulation.
- ArduPilot
- SITL
- Gazebo
- Python
- Problem
- Autonomy code has to survive sensor failures, but testing those failures usually means brittle, one-off scripts.
- Built
- A reusable, plugin-based framework on ArduPilot SITL + Gazebo that injects GPS faults and characterizes airspeed failures, organized around a structured multi-stage attempt lifecycle.
- Result
- Systems-level test architecture, not throwaway scripts — designed to add new fault types as plugins.
-
Flightmare RL — flight-control pipelines
Reinforcement-learning training for quadrotor navigation policies.
- SAC
- PyTorch
- Gymnasium
- sim
- Problem
- Training flight policies needs a clean, reproducible loop between the simulator, the RL algorithm, and evaluation.
- Built
- A modern RL stack (PyTorch + Stable-Baselines3 + Gymnasium) on the Flightmare simulator: a production-ready SAC implementation across hover, target-reaching, and obstacle-avoidance tasks, with vectorized parallel environments and a Docker build system.
- Result
- A clean, typed, tested training pipeline for continuous-control flight policies.
About
Who I am
I am an electronics-engineering graduate from the American University in Cairo (full merit scholarship), now the sole AI engineer at an autonomous-drone startup.
I work across the whole stack of the problem — from RL theory and reward design, to the simulation infrastructure agents train in, to shipping packaged open-source tools.
Before this I did R&D at Siemens EDA on Calibre xACT (parasitic-capacitance extraction) and research at the AUC Center for Nanoelectronics.
Based in Cairo and open to remote EU engineering roles or relocation.
Contact
Get in touch
Open to remote EU roles and relocation.