Looking for a job

majd souissi

@majdsouissi

Message

Applied AI & MLOps engineer building production scoring systems and multimodal RAG pipelines with LLMs.

Tunisia

Message

At Bourse des Crédits, I engineered a hybrid scoring engine that combines business rules, Gradient Boosting, and explainability to prioritize client files and recommend insurance offers. I deployed it as a FastAPI and Docker microservice with MLflow retraining pipelines.

I also integrated SHAP attributions with a locally hosted Qwen 2.5 model to generate French explanations, validated through an automated LLM-as-a-Judge pipeline.

At the Data Engineering & Semantics Research Unit (FSS), I architected a real-time video-indexing pipeline using Apache Kafka and FAISS. I benchmarked quantized vision-language models, with Qwen-VL 2.5 7B achieving 76.4% accuracy.

At LaRINa Research Laboratory, I built and shipped the AlmBark ID mobile application in Flutter, with an on-device vision pipeline. My projects also include a self-correcting multimodal RAG system and an autonomous API and code test-generation agent.

Experience

Work history, roles, and key accomplishments

AI Engineering Intern (End-of-Studies)

Bourse des Crédits

Feb 2026 - Jun 2026 (4 months)

Engineered a hybrid scoring engine combining business rules, Gradient Boosting, and explainability to prioritize client files and recommend insurance offers. Containerized and deployed the scoring microservice via FastAPI and Docker, integrating automated MLflow retraining pipelines.

Python Gradient Boosting SHAP MLFlow fastAPI Docker Qwen2.5 Ollama

AI Engineer Intern

Data Engineering & Semantics Research Unit (FSS)

Aug 2025 - Sep 2025 (1 month)

Architected a real-time video-indexing pipeline using motion detection and keyframe extraction, streaming video payloads via Apache Kafka into FAISS vector indices. Benchmarked and deployed quantized 7B Vision-Language Models on Tesla T4 GPUs.

Kafka Faiss Qwen2.5 VL LLaVA NeXT SmolVLM Model Quantization Tesla T4

Edge AI & Deep Learning Intern

LaRINa Research Laboratory

Jun 2025 - Aug 2025 (2 months)

Audited CNN feature attribution using Grad-CAM and LIME to eliminate background shortcut learning. Built and shipped the AlmBark ID cross-platform mobile application in Flutter, exporting the vision pipeline to PyTorch Lite.

Python PyTorch Lite Tensorflow Lite CNN Grad CAM LIME Flutter Edge AI Android

Education

Degrees, certifications, and relevant coursework

National Higher School of Engineers of Tunis (ENSIT)

National Engineering Degree, Applied Mathematics and Modeling

2023 -

Pursuing a National Engineering Degree in Applied Mathematics and Modeling from September 2023 to September 2026.

National Higher School of Engineers of Tunis (ENSIT)

Research Master's Degree, Data Science

2025 -

Grade: Mention Très Bien / High Honors

Pursuing a Research Master's Degree in Data Science with high honors, from September 2025 to October 2026.

Preparatory Institute for Engineering Studies of Sfax (IPEIS)

Preparatory Cycle, Physics & Chemistry

2021 - 2023

Completed the Preparatory Cycle in Physics & Chemistry from September 2021 to July 2023.

NVIDIA Deep Learning Institute

Certification, Deep Learning

Completed certifications in Deep Learning Fundamentals, Computer Vision for Industrial Inspection, Generative AI with Diffusion Models, and Rapid App Development with LLMs.