AI Research Engineer — Computer Vision
I build vision models that turn raw video into structured understanding of complex activity, most recently in sports broadcasts.
Right now I'm drawn to generative world models: models that learn how a world behaves and generate it frame by frame as you move through it. They meet a question I love reading about, simulation theory, and AI is how I get to explore it. Beyond the screen, I look up: the universe, and what it would take for humanity to climb the Kardashev scale.
PerceptionMotionStructureUnderstanding
Lately I have been designing and training models for sports AI, contributing to racket-sport video analysis in a research program with the French Tennis Federation. Before that, I built computer-vision, NLP and conversational-AI systems in research and industry. Open any work for its architecture, results and what I built.
French Tennis Federation × Efrei Research Lab · 2023 – 2026
01WhatSkeleton-based action recognition
Fédération Française de Tennis
Efrei Research LabParis, FranceRecognises fine-grained strokes from skeleton and ball dynamics.
02HowSelf-supervised representation learning
Fédération Française de TennisEfrei Research LabParis, FranceDiscovers playing styles no one labelled.
03WhenVideo understanding
Fédération Française de TennisEfrei Research LabParis, FranceFinds the play inside hours of broadcast.
04WhereSports field registration
Fédération Française de TennisEfrei Research LabParis, FranceGeometry from a single broadcast frame.
Research and industry internships · 2020 – 2022
05Sports video analysis
Fédération Française de Badminton
Efrei Paris · Efrei Research LabVillejuif, FranceFrom a badminton rally to tactics, automatically.
06NLP · topic modelling
The Story MarketMunich, Germany (remote)A month of German news, organised into topics.
07Conversational AI · NLP
Alight Consulting GmbHOttobrunn, Germany (remote)A learning assistant that understands what you ask.
About
I'm fascinated by how a machine can learn the rules of a world just by watching it. That is where my attention is today: generative world models, models that learn how the visual world evolves and generate the next frames of an explorable world as you move through it.
Those models meet a question I keep reading about: simulation theory. If a model can learn to simulate a world convincingly from pixels alone, what does that say about the worlds we take as real? AI is the way I get to explore that question with my own hands, one model at a time.
Beyond the screen, I look up. I love the universe, its scale and its open questions, and I'm endlessly curious about what it would take for humanity to climb the Kardashev scale, from harnessing the energy of our planet to that of our star. Building more capable, more grounded intelligence feels like one of the steps.
Recommendations
I had the pleasure of working with Ahmed Jouini during a collaborative research initiative between EFREI Research Lab and the French Tennis Federation (FFT).
Throughout this multi-phase research project, Ahmed played a central role in transforming an applied need from elite-level tennis into rigorous scientific work in computer vision, deep learning, and sports analytics. His contributions went beyond the design of an extensive annotated video database spanning several tournament seasons: he formulated research problems, analyzed the state of the art, developed original methods for tennis video understanding, and evaluated them through systematic experiments to generate actionable insights for tennis performance and decision-making. His work led to peer-reviewed international conference publications, reflecting both its scientific quality and its practical relevance for the French Tennis Federation.
Ahmed approaches research with remarkable rigor and intellectual curiosity. He demonstrated a high level of autonomy while consistently delivering high-quality, well-structured contributions. Beyond his technical expertise, what stands out is his ability to connect complex AI concepts with real-world challenges in elite-level tennis, bringing both scientific depth and strong business understanding.
Ahmed is a passionate and committed researcher with a rare combination of technical excellence and strategic vision. I highly recommend him to any organization seeking a talented AI scientist capable of driving impactful innovation.
Étienne PernotDirecteur de la Recherche et de la Valorisation · EFREI Paris
Ahmed is a PhD student in Efrei Research Lab, arriving soon at the end of this curriculum.
Beyond his impressive skills and autonomy in video analysis, it is his ability to interface with experts from the French Tennis Federation that is truly remarkable. Working in interdisciplinary way may seem easy, but it requires respect, active listening, and the energy to understand the other discipline, its expectations, and its constraints. The goal is indeed to put digital technology at the service of other sciences, not to impose digital domain's vision and constraints. Ahmed deeply understand and has applied that successfully, with humility and exceptional proficiency.
Letter of recommendation
From its inception to the evaluation phase, Mr Ahmed Jouini served as Project Lead [2023 – 2026] and has been a central figure in the design, execution, and delivery of this ambitious and technically complex program.
I strongly and unreservedly recommend Mr Ahmed Jouini for any research or professional position requiring excellence, innovation, and commitment.
Alex LothCoordonnateur Recherche et Innovation · Direction des Systèmes d'Information, Fédération Française de Tennis (FFT)
Contact
I'm looking for AI research engineer roles in computer vision and video understanding, ideally in sports analytics, though not only. Ready to start a new journey.
Paris, France