I am a PhD researcher at CentraleSupélec / Université Paris-Saclay, conducting my research within ONERA / SONDRA, a France–Singapore research alliance involving CentraleSupélec, ONERA, NUS, and DSO National Laboratories. My work focuses on robust radar target detection, generative modeling, out-of-distribution detection, and complex-valued learning. In parallel, I build reproducible software and experimental pipelines shaped by scientific-software experience in testing, documentation, packaging, CI/CD, and deployment-oriented engineering. I am seeking roles in quantitative research / quantitative ML, AI research and ML engineering, and defense AI, with a particular interest in mathematically grounded, high-impact environments where research originality and rigorous experimental execution both matter.
I position my profile across three complementary tracks, united by the same core strengths: mathematical grounding, research originality, and dependable engineering execution.
Statistical learning, robust detection under severe noise, probabilistic modeling, and structured signal reasoning, extended to market microstructure, transaction-cost-aware experimentation, and short-horizon alpha research. I am particularly interested in roles where mathematical modeling, data-driven signal design, and rigorous benchmark construction matter as much as raw predictive performance.
Research and engineering around diffusion models, VAEs, OOD detection, evaluation, and reliable ML systems. I do not only implement ideas from the literature: I also formulate new research directions, build them into code, and validate them through ablations, benchmarks, and rigorous experimental analysis.
End-to-end research experience in robust radar target detection, complex-valued deep learning, whitening and detector-fusion strategies, and statistical reasoning under challenging clutter and thermal-noise conditions.
A view of the themes I work on, the problems I solve, and the current maturity of each line of work.
My PhD work focuses on out-of-distribution radar detection. I design original methods and evaluation protocols around real-valued and complex-valued VAEs, latent-space metrics, whitening strategies, and fusion with classical detectors, and I am extending this research to diffusion and flow-based approaches.
At PHIMECA, I worked on uncertainty quantification and scientific software engineering, including testing, documentation, packaging, CI/CD integration, and executable delivery. This experience shaped the way I build research code: structured, reproducible, and usable beyond a single experiment.
This research direction studies how score-based and diffusion models can support training-free probing, posterior-consistency analysis, score-space geometry, and interpretable anomaly localization. It currently includes GEPC and follow-on work aimed at developing new OOD methods across both image and signal settings.
A structured research platform for limit-order-book modeling, execution benchmarking, real walk-forward alpha evaluation, and robustness under market shift. The project combines microstructure features such as spread, imbalance, microprice, and signed flow with explicit transaction-cost modeling and reproducible benchmarking across four phases: execution, controlled alpha sandbox, real-data alpha, and cross-venue robustness.
Across research and software work, I operate from method design to implementation, evaluation, documentation, testing, CI/CD, and packaging. That combination matters for teams that need original ideas together with dependable execution, whether in AI labs, quantitative research groups, or deep-tech and defense environments.
Public PDF and arXiv links are listed whenever they are available.
Y. A. Rouzoumka, J. Pinsolle, E. Terreaux, C. Morisseau, J.-P. Ovarlez, C. Ren — submitted to ICML 2026 · arXiv preprint
Y. A. Rouzoumka, E. Terreaux, C. Morisseau, J.-P. Ovarlez, C. Ren — ICASSP 2025
Y. A. Rouzoumka, E. Terreaux, C. Morisseau, J.-P. Ovarlez, C. Ren — EUSIPCO 2025
Y. A. Rouzoumka, E. Terreaux, C. Morisseau, J.-P. Ovarlez, C. Ren — ICASSP 2026
J. Pinsolle, Y. A. Rouzoumka, C. Ren, C. Morisseau, J.-P. Ovarlez — ICASSP 2026
Y. A. Rouzoumka, J. Pinsolle, E. Terreaux, C. Morisseau, J.-P. Ovarlez, C. Ren — submitted to IEEE Transactions on Signal Processing · arXiv preprint
Y. A. Rouzoumka, J. Pinsolle, E. Terreaux, C. Morisseau, J.-P. Ovarlez, C. Ren — submitted to EUSIPCO 2026 · arXiv preprint
Y. Rouzoumka, E. Terreaux, C. Morisseau, J.-P. Ovarlez, C. Ren — GRETSI 2025
P. Meena, Y. A. Rouzoumka, J. Pinsolle, C. Ren, M. N. El Korso, J.-P. Ovarlez — submitted to EUSIPCO 2026
Y. A. Rouzoumka, E. Terreaux, C. Morisseau, J.-P. Ovarlez, C. Ren — submitted to JDS 2026
CentraleSupélec · Université Paris-Saclay · ONERA / SONDRA · since October 2023. Research conducted within SONDRA, the France–Singapore research alliance in radar, electromagnetics, signal processing, and AI.
Internship and apprenticeship experience · 2022
IUT Computer Science · Université Paris-Saclay · since October 2024
I am open to conversations about quantitative research, AI research / ML engineering, deep-tech ML, and defense AI roles.