Skip to content
View brahim-elkhobzi's full-sized avatar

Block or report brahim-elkhobzi

Block user

Prevent this user from interacting with your repositories and sending you notifications. Learn more about blocking users.

You must be logged in to block users.

Maximum 250 characters. Please don’t include any personal information such as legal names or email addresses. Markdown is supported. This note will only be visible to you.
Report abuse

Contact GitHub support about this user’s behavior. Learn more about reporting abuse.

Report abuse
brahim-elkhobzi/README.md

Hi, I'm Brahim EL KHOBIZI 👋

AI Full-Stack Engineer | LLM Fine-Tuning & Multi-Agent Systems

📍 Rabat, Morocco


🚀 About Me

I'm an AI Full-Stack Engineer with 3 years of combined experience spanning LLM fine-tuning, RAG pipelines, multi-agent orchestration, and full-stack web development. I graduated with a Master's in AI & Virtual Reality from Université Ibn Tofail, where my thesis focused on frugal, adaptive structured-data generation with LLMs.

  • 🔬 Currently researching efficient fine-tuning (Unsloth, LoRA) and model compression (quantization, pruning) for my Master's thesis
  • 🤖 I build multi-agent systems with LangGraph, orchestrating specialized models that communicate and split tasks
  • 📚 I care about frugal AI: evaluating models not just on accuracy, but on energy consumption and latency
  • 🧩 Background in RAG (retrieval-augmented generation), vector search, and continual/incremental fine-tuning
  • 🌐 Trilingual: French (fluent), English (C1 scientific), Arabic (native)

🛠️ Tech Stack

AI / ML PyTorch Hugging Face LangChain LangGraph FAISS Unsloth

Full-Stack Development Python C# .NET MVC Blazor React Next.js Django Flask

Infra Docker Git


📌 Featured Projects

🔹 TamTrack — Adaptive Multi-Agent Platform Python · LangGraph · Unsloth · Docker | 2025–2026 Orchestrates several specialized agents (course generation, evaluation, feedback) with task distribution between them. Uses incremental fine-tuning to personalize agents per user, with latency and memory footprint tracked as frugality metrics.

🔹 Legal RAG Assistant Python · Llama 3 · LangChain · FAISS | 2024 A retrieval-augmented legal assistant with a dynamically updatable knowledge base — new legal texts can be added without retraining the whole pipeline.

🔹 Lightweight Pictogram Model Prototype PyTorch · Hugging Face · Symbolic rules | 2025 Compressed a DistilBERT model for inference on lightweight/edge terminals, combining neural compression with symbolic rules.


💼 Experience

AI Full-Stack Engineer — Educational Platform · Jan 2024 – Jan 2025 Fine-tuned Llama 3 with Unsloth for the education domain with memory optimization, and built a FAISS-based RAG pipeline to reduce response latency.

C# Developer — Freelance Web Applications · Mar 2023 – Feb 2024 Object-oriented design, unit testing, agile project management.

Data Scientist — Freelance, Optimization · Sep 2022 – Aug 2023 Optimization algorithms and data analysis.


🎓 Education

  • Master's, Artificial Intelligence & Virtual Reality — Université Ibn Tofail, Kénitra (2023–2026) Thesis: Adaptive generation of structured data by LLMs — frugality and evaluation
  • Bachelor's, Mathematics & Computer Science — Université Ibn Tofail, Kénitra (2018–2023)

📊 GitHub Stats


📫 Reach me at brahimkhoobizi21@gmail.com or on LinkedIn

Popular repositories Loading

  1. mini-project-python mini-project-python Public

  2. app app Public

    Python

  3. Cours-Py-AI Cours-Py-AI Public

    Python

  4. app-java-mongoDB app-java-mongoDB Public

    Java

  5. detection-spam-ML detection-spam-ML Public

    Jupyter Notebook

  6. APP-Bank APP-Bank Public

    Python