Building intelligent systems and scalable software through code, machine learning, and deep learning.
02. ABOUT ME
I'm Hamza Lafsioui, an AI Engineer & Full-Stack Developer based in Morocco, currently honing my craft at YouCode (UM6P).
With a background in Physics and Electronics from Université Ibn Zohr, I approach AI problems with strong mathematical discipline, analytical rigor, and systems-level thinking.
My philosophy is straightforward: real-world AI is only as good as the software delivering it. I engineer end-to-end data pipelines, evaluate models with strict metrics, package inference services with FastAPI & Docker, and integrate intelligent capabilities into clean, resilient applications.
During my internship at INVENEO, I developed a mobile recruitment application in React Native and integrated AI models for automated resume extraction and candidate criteria parsing.
+20
Projects Built
95.37%
CNN Model Accuracy
Web & AI
Core Domains
UM6P
Academic Rigor
Specialized training at YouCode (UM6P): computer vision, predictive models, MLflow tracking, and Airflow orchestration.
Designing clean REST APIs, multi-tenant SaaS platforms, and responsive interfaces with React, Next.js, and Laravel.
Building RAG systems for private knowledge retrieval, multi-agent networks with LangGraph, and MCP protocols.
Bachelor’s degree in Physics and Electronics from Ibn Zohr: strong foundation in mathematical analysis and signal processing.
03. TECHNICAL SKILLS
A comprehensive toolkit spanning Machine Learning, Deep Learning, MLOps orchestration, and Full-Stack development.
04. FEATURED PROJECTS
Real intelligent systems with concrete metrics, clean architectures, containerized deployments, and production reliability.
Deep Learning Medical Imaging & Brain Diagnostic Suite
Advanced deep learning and computer vision diagnostic platform for autonomous MRI brain scan analysis. Implements InceptionV3 and YOLOv8 architectures to classify and localize scans across tumor pathologies, featuring image preprocessing and low-latency inference.
MLOps Diabetes Risk & Patient Segmentation Pipeline
Patient clustering with K-Means followed by benchmark evaluation of KNN, SVM, Logistic Regression, and Random Forest on F1-score. Automated model tracking with MLflow, scheduled weekly retraining via Airflow, and FastAPI serving with Streamlit.
Food Delivery Arrival Prediction Engine
Machine learning regression pipeline for accurate delivery arrival time estimation. Comprehensive data cleaning, exploratory data analysis, and advanced geospatial feature engineering using Haversine distance.
Meteorological Data Pipeline & Medallion Architecture
Automated data engineering pipeline extracting real-time meteorological metrics across 120 Moroccan cities via Open-Meteo API. Built with Medallion Architecture (Bronze/Silver/Gold), orchestrated with Apache Airflow, PostgreSQL data warehouse, and interactive Streamlit analytics.
Football Match Ticketing System & Digital QR Entry
Premium football match ticketing platform built with PHP and Tailwind CSS. Features digital ticket issuance with anti-fraud QR codes for automated stadium check-in, tiered seating pricing, real-time sales & revenue analytics, and multi-role dashboards for buyers, organizers, and administrators.
Real-Time Job Marketplace & WebSocket Messaging
Full-stack containerized real-time job search platform. Built with a robust Laravel REST API backend, reactive React & Vite frontend, instant messaging via Laravel Reverb WebSockets, and asynchronous Redis queue workers.
Intelligent E-Learning Platform with Stripe & Recommendations
Comprehensive e-learning platform enabling students to explore curated courses, manage favorites, and receive personalized recommendations. Features secure role-based REST APIs with Laravel 12 and automated Stripe checkout.
Smart Shared Housing & Debt Settlement Platform
Collaborative shared living platform built with Laravel 12. Implements a minimum-transaction debt settlement algorithm to simplify shared expenses, token-based invitations, budget caps, and tenant reputation scoring with Pest PHP testing.
Multi-Tenant SaaS Car Rental Platform
Enterprise-grade full-stack multi-tenant car rental SaaS platform architected using Clean Architecture. Enables independent tenant isolation where each tenant administers multiple vehicle agencies with dedicated subdomains.
High-Performance Clean Architecture CQRS API
Secure RESTful microservice API with JWT authentication and granular Role-Based Access Control (Admin, Speaker, Attendee). Designed with Clean Architecture and CQRS pattern using MediatR, distributed Redis caching, and Serilog logging.
Driver & Vehicle Licensing Desktop Suite
Comprehensive desktop application managing the full lifecycle of driving licenses and vehicle registrations. Built with N-Tier architecture and ADO.NET for high-throughput transaction execution and automated administrative PDF reporting.
05. EXPERIENCE & EDUCATION
From physics and electronics to AI engineering at YouCode (UM6P) and real-world industrial software delivery.
Intensive project-based curriculum covering Machine Learning pipelines, Deep Learning, MLOps, LLMs/RAG, and enterprise Full-Stack development.
Rigorous foundation in applied mathematics, signal processing, electromagnetism, and embedded electronic systems.
High school diploma with a scientific concentration in physics and mathematics.
06. CONTINUOUS EVOLUTION
The AI frontier is moving fast. Here are the specialized domains, architectures, and tools I am actively exploring.
Designing stateful, multi-agent cooperative networks that reason, execute dynamic tool calls, loop over validations, and self-correct using LangGraph and Python.
Building hybrid search retrieval (BM25 + Dense embeddings), multi-stage reranking (Cohere), and chunking optimization for zero-hallucination enterprise QA.
Implementing open standard MCP servers and clients to bridge language models directly with enterprise repositories, operational databases, and tool environments.
Automated drift detection, data versioning with DVC, CI/CD with GitHub Actions, Langfuse observability for LLMs, and zero-downtime model deployments.
Enforcing strict schema validation (Pydantic), hallucination benchmarks (Ragas, DeepEval), jailbreak defense, and automated output sanitization.
Evaluating low-latency inference serving with vLLM, TensorRT-LLM, quantization techniques (AWQ, GGUF), and deploying scalable endpoints on cloud infrastructure.
07. INITIATE CONTACT
Whether you are looking to integrate machine learning into your infrastructure, discuss AI roles, or collaborate on a project.
Ouled Teima / Safi / Marrakech, Morocco
Available for on-site & remote positions