MSc student in Control, Signal & Image Processing (M2 ATSI) · Université Paris-Saclay
Computer Vision · Inverse Problems & Imaging · Deep Learning · Real-time Control
I design algorithms that turn raw measurements into reliable decisions: reconstructing images from blurred, noisy data, detecting and classifying objects, and controlling physical systems in real time.
- 🎓 M2 ATSI (Automatique, Traitement du Signal et des Images), Université Paris-Saclay, with CentraleSupélec, IOGS and ENS Paris-Saclay
- 🔬 Research project (TER): Bayesian image deconvolution with automatic hyperparameter estimation, compared with Diffusion Posterior Sampling (diffusion-model priors)
- 🎯 Looking for a 6-month internship (March → September 2027) in computer vision, medical imaging, perception / ADAS or signal processing, ideally on real data rather than toy datasets
- 🌍 French (fluent) · English (professional working proficiency)
Research project (TER), Université Paris-Saclay · supervised by François Orieux · 50-page report
- Recover a sharp image from a blurred, noisy observation
$z = Hx + b$ (an ill-posed inverse problem) - Wiener-Hunt quadratic regularization solved in closed form with FFT (circulant approximation, $O(N \log N)$)
- Unsupervised hyperparameter estimation by marginal likelihood: the likelihood minimum matches the MSE minimum, so no manual tuning is needed
- Diffusion Posterior Sampling (Chung et al., 2023) reproduced in PyTorch: DDPM + U-Net prior, 1000-step guided sampling
- Result: Wiener-Hunt 25.68 dB vs DPS 24.46 dB PSNR (degraded input: 23.51 dB). DPS gives sharper edges, but a model-based method still wins when the learned prior has seen a single image
Python PyTorch NumPy SciPy FFT Bayesian inference
| Project | What I did | Stack |
|---|---|---|
| Sliding-mode control of a 3-DOF robotic arm | Dynamic model, SMC law design, real-time implementation on dSPACE; lower tracking error than a PID baseline | MATLAB/Simulink, dSPACE, IMU |
| Vision pipeline | Real-time object detection and image processing | YOLOv8, OpenCV |
| Predictive maintenance of an electric motor | Sensor data acquisition, failure prediction, LSTM vs Random Forest comparison, real-time monitoring | Python, LSTM, Random Forest, Raspberry Pi |
| Image processing & machine learning | Filtering, edge detection, thresholding; image classification with SVM and CNN | Python, OpenCV, scikit-learn |
| Advanced signal processing | Matched filtering and detection in noise, time-delay estimation, AR modelling (Yule-Walker), source localization (MUSIC, maximum likelihood) | MATLAB |
| Gesture-controlled robotic arm | Hand-gesture recognition to drive a pick-and-place arm | Python, OpenCV |
| DC motor speed & position control | Modelling and controller design | MATLAB/Simulink |
Languages
Computer vision & deep learning
Control & embedded
Methods
- Signal & image: filtering, detection, spectral estimation, restoration, segmentation, Markov random fields
- Inverse problems: deconvolution, variational & Bayesian regularization, proximal algorithms, diffusion priors
- Estimation: Kalman filtering, maximum likelihood, Cramér-Rao bound, system identification
- Control: PID, RST, state feedback, LQR/LQG, H∞, MPC, sliding-mode control
FTTH engineering intern · KABTELECOM (partner of SOGETREL for Orange) · Normandy · May–Jul 2025 Fiber installation, splicing and commissioning; OTDR and power-meter testing, RSSI/throughput diagnostics; support for telecom project planning.
Electronics / automation maintenance intern · Sarl Ibrahim & Fils (Ifri) · Béjaïa, Algeria · Jun–Jul 2024 Production-line analysis, Siemens S7-1200 PLCs with TIA Portal, preventive-maintenance diagnostics.
Organizing member & mentor · IEEE Student Branch, University of Boumerdès · 2021–2022
| Degree | Institution | |
|---|---|---|
| 2026–2027 | M2 ATSI: Control, Signal & Image Processing | Université Paris-Saclay |
| 2025–2026 | M1 E3A, ASDS track: Control, Data Science, Signal | Université Paris-Saclay |
| 2024–2025 | Licence SPI: Electronics, Signal & Networks | Université Sorbonne Paris Nord |
| 2023–2024 | M1 Electrical Engineering | IGEE (ex-INELEC), Algeria |
| 2020–2023 | Licence (BSc) Electrical & Electronics Engineering | IGEE (ex-INELEC), Algeria |
Key M2 courses
Scientific Computing in Python · Estimation & Identification (Kalman, ML, Cramér-Rao) · Optimization (duality, proximal algorithms) · Statistical & Reinforcement Learning · Multivariable Control (LQR, LQG, H∞, μ-analysis) · Signal Processing for Imaging Systems · Machine & Deep Learning · Model Predictive Control · Advanced Image Processing (restoration, segmentation, MRF) · Inverse Problems (variational & Bayesian)
🇫🇷 En français
Étudiant en Master 2 ATSI (Automatique, Traitement du Signal et des Images) à l'Université Paris-Saclay. Je recherche un stage de fin d'études de 6 mois à partir de mars 2027 en vision par ordinateur, imagerie, perception ou traitement du signal.
Mon TER porte sur la déconvolution d'images : régularisation de Wiener-Hunt avec estimation automatique des hyperparamètres par inférence bayésienne, comparée à un a priori appris par modèle de diffusion (DPS). J'ai aussi implémenté une commande par modes glissants en temps réel sur dSPACE, un pipeline de vision YOLOv8/OpenCV et un système de maintenance prédictive (LSTM / Random Forest).
♟️ Chess (1800 Elo) · ⚽ Football & running ·