AI ENGINEER · TUNISIA
Turning messy problems
into working systems.
I design and build AI systems, data pipelines, and tools that solve real-world problems — from messy data and disconnected systems to usable products people actually rely on.

- 8 systems → 1 CleverTap view
- 98–100% record coverage
- Hours → seconds for AI action plans
[ EXPERIENCE ]
Where I've built.
Poulina Group Holding · AI Engineer
Oct 2025 ––
Building production AI and data systems for hotels and parks. Delivered a multi-hotel complaint analytics platform with LLM classification and automated action plans, and a CleverTap integration unifying 8 systems across 5 hotels and 2 parks with hourly sync.
- FastAPI
- React
- LLaMA
- Groq
- ETL
- Docker
BinitNS (Accelex) · AI Engineer Intern
Feb 2025 – Jun 2025
Deployed Qwen2.5-VL on AWS SageMaker via Ollama for real-time document parsing, and proposed a VLM-based replacement for a legacy Tesseract OCR pipeline.
- OCR
- Qwen2.5-VL
- SageMaker
- Ollama
Approach AI · AI Engineer Intern, LegalBot
Oct 2024 – Feb 2025
Built a legal document preprocessing and OCR pipeline on Azure, plus a FAISS-based RAG, and cut Q&A pair generation time by 90%.
- Azure ML
- GPT-4
- Llama 3
- FAISS
SACEM Industries · AI Engineer Intern, HR chatbot
Jun 2024 – Aug 2024
French-language HR chatbot: LLaMA3 8B fine-tuned with LoRA, RAG with FAISS, Streamlit interface.
- LoRA
- RAG
- FAISS
- Streamlit
Omnilink · AI Engineer Intern, speaker identification
Jun 2024 – Aug 2024
Speaker diarization and transcription with pyannote and Whisper; LLaMA2 classifies speakers as employee or customer.
- pyannote
- Whisper
- LLaMA2
- Flask
Tunisie Telecom · ML Intern, SMS spam detection
Jul 2023 – Aug 2023
Compared Multinomial Naive Bayes, XGBoost, and fine-tuned BERT for spam detection.
- NLP
- XGBoost
- BERT
[ SELECTED PROJECTS ]
A few things I've built.
Real problems, practical engineering, and systems designed to work beyond the demo.
Complaint Analytics Platform
Turning hotel complaints into structured insights and actionable plans.
- FastAPI
- React
- LLaMA
- Groq
- Docker
CleverTap Integration
Eight systems. Five hotels. Two parks. One reliable customer view.
- Python
- ETL
- CleverTap
- Docker
- Data Quality
OCR Pipeline Upgrade
Making document understanding more practical for real-world workflows.
- OCR
- Qwen2.5-VL
- AWS SageMaker
- Ollama
- Python
[ ON GITHUB ]
- under-the-hood01
Visual, step-by-step explainer for all 150 NeetCode problems — animated walkthroughs built with Django.
- Python
- char-transformer02
A GPT-style character-level transformer language model.
- Python
- ai-ml-flashcards03
Flashcards for AI and machine learning concepts.
- HTML
[ SKILLS ]
- AI/ML
- PyTorch
- TensorFlow
- Scikit-learn
- Hugging Face
- LLMs and NLP
- Transformers
- SpaCy
- LLaMA
- GPT-4
- BERT
- LoRA
- RAG
- FAISS
- Backend
- Python
- FastAPI
- Flask
- React
- Cloud and DevOps
- AWS SageMaker
- Azure ML
- Docker
- Nginx
- Git
- Languages
- Arabic (native)
- French (DELF B2)
- English (TOEIC)
[ EDUCATION ]
Engineering degree, Statistics and Data Analysis
Higher School of Statistics and Information Analysis
2022 – 2025
Pre-engineering (Maths and Physics)
Preparatory Institute of Nabeul
2020 – 2022
[ ABOUT ]
An engineer who likes building.
Based in Tunisia
I'm Hamza, an AI engineer based in Tunisia. I work at the intersection of AI, data, and software engineering, with a focus on turning difficult, messy problems into systems people can actually use.
I enjoy owning a problem from understanding the workflow to building, deploying, and improving the solution. For me, a model is only one part of the system. Reliability, data quality, usability, and measurable impact matter just as much.