Torat / Afric'Artech
AI-powered platform for Moroccan artisans
Built and shipped a complete AI-powered product for Moroccan artisans to help them sell online during a 48-hour hackathon.
From interface to infrastructure.
I build custom websites and e-commerce platforms, Android and iOS apps, desktop software, and AI integrations. I also optimize existing code and websites for speed, reliability, and maintainability.
Engineered from interface to deployment: local AI assistants, automated pipelines, and full-stack platforms.
Local-first AI personal assistant
A fully local, multi-LLM AI desktop assistant with voice interaction, semantic memory, and a modular skills system.
Local-first architecture with GPU acceleration, vector database for semantic memory, multi-LLM orchestration, and modular skills system.
AI-powered platform for Moroccan artisans
Built and shipped a complete AI-powered product for Moroccan artisans to help them sell online during a 48-hour hackathon.
End-to-end AI content generation system
A complete local AI pipeline for automated content generation using n8n, Ollama, ComfyUI, and FLUX.
Full-stack blogging platform
A complete blogging platform built with Java Spring Boot backend and Angular frontend with JWT authentication.
Offline-resilient mobile application
A mobile application with offline-resilient architecture, watcher dashboard, and AI assistant built with React Native and Go.
Full-stack social platform
A complete social networking platform with real-time features, WebSockets, and optimized database performance.
HTTP/1.1 server in Java
A custom HTTP/1.1 server built in Java on non-blocking NIO. A single event-driven selector loop handles multiple ports, request routing, GET/POST/DELETE, file uploads, cookies and sessions, configurable error pages, and CGI execution.
faster-whisper / CUDA
Python / LangChain / Router
Ollama (Llama 3 / Mistral) & Claude 3.5 API
ChromaDB / all-MiniLM-L6-v2
n8n / Bash / Python Tool Modules
Kokoro TTS / 82M PyTorch
Technical proficiencies across the entire engineering spectrum: client interfaces, distributed backends, local AI pipelines, and Linux security.
Professional roles, intense 42-style peer-learning curriculum at Zone01 Oujda, and 60+ rigorous code reviews.
Leading product engineering and delivery across mobile, desktop, and web applications, with AI capabilities integrated into the products.
Evaluate and filter multimodal AI responses to support quality, relevance, and safety across text, images, video, and voice.
Building production web applications, AI systems, and automation infrastructure for clients across various industries.
42 School-style intensive, project-based peer-learning program focusing on algorithms, data structures, full-stack development, Docker, and cybersecurity.
Kernel tuning, systemd process topologies, least-privilege PAM isolation, and reproducible Infrastructure as Code.
Debian server setup with LVM, static networking, systemd services, and operational runbooks.
In-depth documentation on local LLMs, CUDA tensor acceleration, vector RAG retrieval, and real-time audio systems.
A comprehensive guide to building a local AI assistant using Ollama, including setup, configuration, and best practices.
Understanding Retrieval-Augmented Generation implementation using ChromaDB for semantic memory in AI applications.
Practical experience running large language models locally on consumer hardware with CUDA optimization.
Creating a complete voice interaction pipeline using faster-whisper for speech-to-text and TTS for response.
Architecting a system that orchestrates multiple LLM providers (Ollama, Claude, Groq) for optimal performance.
Available for select full-stack roles, local AI infrastructure contracts, and systems auditing.
mehdimoulabbi1@gmail.com
github.com/M-MDI
linkedin.com/in/mmoulabbi