MG.
Building at Paymob

02 / Projects

Built to be useful.

Retrieval systems, language models, AI agents, and computer vision. A selection of the things I build to connect ideas with real problems.

01 / Build archive

Systems. Research. Experiments.

4 projects
INDEX → RETRIEVE → GENERATE
≡Documents→RAG→✧Answers
RELEVANT CONTEXT. USEFUL ANSWERS.

001 / Retrieval systems

Mini-RAG

A modular system for document indexing, vector search, and answers grounded in relevant context.

FastAPIQdrantDocker
GitHub ↗
AI-MUFASSIR / MODEL ACCURACY
65%→83%+
FINE-TUNING + RETRIEVAL

002 / Language & research

AI-Mufassir

Quran question answering with fine-tuned transformers and retrieval-augmented generation. Accuracy improved from 65% to over 83%.

TransformersLangChainRAG
GitHub ↗
YOUR PERSONAL STUDY CREW
Read01Write02Listen03Speak04
FOUR SKILLS. ONE PLAN.

003 / AI agents / education

IELTS / TOEFL Coach

An AI study assistant that creates personal preparation plans and gives writing feedback through a chat interface.

CrewAIStreamlit
GitHub ↗
COMPUTER VISION / MODEL REFINEMENT
98%
FROM A 97% BASELINE

004 / Computer vision

Face Recognition

A convolutional neural network pipeline refined through dropout, data augmentation, and systematic tuning. Accuracy increased from 97% to 98%.

CNNsOpenCVDeep learning
GitHub ↗

02 / Interactive study

From context to clarity.

retrieval-walkthrough / local demo

Choose a question and follow the retrieval process.

  1. 01 / Understand the questionReady
  2. 02 / Find relevant contextWaiting
  3. 03 / Connect the dotsWaiting
Select a question to explore how retrieval works.

Guided example using my portfolio. No live AI model.

↗ / Start a conversation

Have a good question?
Let’s build on it.

Let’s talk ↗

Approach & implementation

View on GitHub ↗