Artificial Intelligence Consultant
Helping companies and enterprises adopt practical, private, and production-ready AI
I advise companies and enterprises on AI strategy, architecture, adoption, and delivery. I help decision-makers identify high-value opportunities, assess technical feasibility, manage risk, and turn AI initiatives into secure, production-ready solutions.
My consulting focus is private, offline AI: local LLMs and RAG, multilingual speech pipelines (STT → translation → TTS), OCR, edge inference, computer vision, and robotics. I evaluate each solution against business value, performance, security, maintainability, and total cost of ownership.
- 🧭 AI strategy and roadmaps aligned with measurable business outcomes
- 🏢 Enterprise AI adoption, from readiness assessment to production planning
- 🔒 Private and on-premises AI for organizations with strict security and data-sovereignty requirements
- 🧠 Solution architecture for local LLMs, RAG, OCR, speech, computer vision, and robotics
- ⚖️ Technology and vendor evaluation based on capability, risk, maintainability, and total cost of ownership
- 🤝 Executive and technical advisory that connects business leadership with delivery teams
These games give robots and LLM agents a place to learn and play.
Each project is a self-contained 3D environment with a machine-facing control API. An agent can observe the state, query available actions, and act. This makes the games useful for testing perception, decision-making, and task execution in software before moving to physical hardware. Humans can play too, but they are the secondary audience.
| Project | Environment | Agent interface | Built with |
|---|---|---|---|
| quantum-chess-game | Offline 3D chess for humans, Stockfish, robots, and LLMs | Observe board · list legal moves · play a move | Three.js · chess.js · Stockfish · Vite · Tauri |
| quantum-garden-game | Offline 3D garden cleanup and tree planting; seeded levels, single/two-player, Arabic + multilingual UI | Model-driven control API for robot/agent skill learning | JavaScript · Three.js · Vite · Tauri |
Both run fully offline and support browser, desktop (Tauri), and mobile targets.
| Project | What it is | Built with |
|---|---|---|
| live-player ⭐ 10 | Resilient Android background audio streaming that handles crashes, network loss, reboots, and battery optimization | Kotlin · Media3/ExoPlayer |
| nebulaterm-ai | AI-powered SSH and terminal client — native terminal, multi-tab sessions, command generation and error diagnosis | Rust · JavaScript · SSH |
| Maya-iptv | One IPTV codebase across web, React Native, Android, iOS, Electron, webOS and Tauri | TypeScript · React Native |
| social-media-saver | Tool for downloading and archiving social media content | TypeScript |
| islamic-loan | Sharia-compliant loan/financing calculator | HTML · JavaScript |
| Diffscript | CLI utility for creating filtered deployment diffs from files changed since a given date | Python |
Groundwork for the offline voice pipeline:
| Project | What it is | Built with |
|---|---|---|
| RNNoiseCancellation | RNN-based real-time noise suppression for speech capture | Python |
| webrtc-aec | WebRTC acoustic echo cancellation for clean microphone input while a device is speaking | Python |
| Habibi-TTS | Unified-dialectal Arabic speech synthesis, tracked and adapted for Arabic TTS research | Python |
| Package | What it is | Downloads |
|---|---|---|
| @codejq/react-simple-snackbar | Modernized React snackbar with React 16.8–19 support, TypeScript types, and flexible positioning | |
| EPPlus.Core.CoreCompat | Cross-platform Excel generation and reading on modern .NET runtimes | |
| ImageProcessor.Core.CoreCompat | Fluent image processing — resize, crop, filters, watermarking, format conversion |
Also maintained or adapted:
- OpenFace-3.0 — facial landmarks, action units, eye gaze, and emotion recognition
- Common-PrintServer — direct browser-to-printer server
- go-whatsapp-web-multidevice — WhatsApp multi-device gateway
- Offline AI voice translation for a government agency — fully on-premises speech recognition, translation, and voice output; no sensitive data leaves the network
- Modular multi-tenant ERP — accounting, invoicing, inventory, assets, projects, HR/payroll, clinics and configurable industry modules
- Enterprise RAG + OCR knowledge platform — Arabic/English OCR, multilingual embeddings, Qdrant vector search, locally hosted models, retryable ingestion
- National training portals — public-sector learning platforms serving 10,000+ users across authentication, courses, enrollment, attendance, and reporting
- Computer vision & access control — facial recognition, biometrics, and camera integrations tuned for real-time production accuracy
- Performance engineering — redesigned a MongoDB aggregation over ~10M documents from ~43 minutes to ~46 seconds
- M. A. El-Sayed, A. B. Ibrahim — Gender Identification Using 3D Touch Sensor, IEEE 3ICT 2018.
10.1109/3ICT.2018.8855776
- MSc Computer Science — Optics, Image, Vision & Multimedia · Université Paris-Est Créteil (UPEC), France
- BSc Physics — Minya University, Egypt
- Professional Scrum Product Owner · Deep Learning AI · Google Project Management Professional
Available for AI consulting and enterprise advisory engagements.