CV | Noé Lallouet

Education

PhD in Artificial Intelligence
Joint research agreement between PSL University and Thales under the supervision of Prof. Tristan Cazenave.
Research topic: Optimizing the trade-off between neural network performance and complexity for radar target detection.

Master’s degree – Artificial Intelligence, Systems, Data (IASD)
Graduated with 1st class honours (15.95/20) and ranked 2nd out of 25 students.
Relevant coursework: Deep learning, NLP, Optimization, Graph mining, Machine learning, etc.
Master’s thesis: Deep learning applied to radar target detection.

Six-month research internship at the Adobe Research Paris lab, focusing on generative modeling and 3D representation learning for Gaussian splatting interactive applications.
Internship topic: Reference-Free Gaussian Splatting Inpainting with 3D-Native Generative Priors.
Responsible for leading the research project, designing and performing key experiments using large-scale computing infrastructure.
Collaboration with UK and US-based research teams.

Responsible for investigating resource-efficient neural network design for the problem of airborne radar target detection.
Developed solutions for designing efficient convolutional neural networks in resource-constrained contexts for implementation in operational radar systems.
Designed neural radar detectors reducing inference latency by 76% while improving long-range detection probability by 5%, enabling deployment in critical embedded hardware.
Scientific contributions to AI-driven radar signal processing and to multi-objective optimisation, illustrated by first-author papers at international conferences.

Responsible for the development and implementation of deep learning algorithms applied to radar target detection.
Trained CNNs for target detection on extremely low SNRs (signal-noise ratio).
Improved the technology readiness level of the developed algorithm towards operational deployment.

Co-founded a R&D-oriented startup bringing machine learning innovation to the field of indoor boulder climbing.
Responsible for the development, training and deployment of state‑of‑the‑art deep learning algorithms in object detection, sequence generation, autoencoding, image generation.
Raised 60k€ pre-seed funding for a 1M€ valuation.

Noé Lallouet, Tristan Cazenave, Cyrille Enderli and Léo Monnier. Multi-Objective Architecture Search for Real-Time Radar Target Detection on Embedded Systems. 2026 International Radar Symposium (IRS), IEEE, https://doi.org/10.23919/IRS70539.2026.11549203
Noé Lallouet, Tristan Cazenave and Cyrille Enderli. “Pareto‑NRPA: A Novel Monte‑Carlo Search Algorithm for Multi‑Objective Optimization.” ECAI 2025, IOS Press, https://doi.org/10.3233/faia251394.
Noé Lallouet, Tristan Cazenave, Cyrille Enderli and Stéphanie Gourdin. “Searching Efficient Deep Architectures for Radar Target Detection Using Monte‑Carlo Tree Search.” 2024 International Radar Conference (RADAR), IEEE, https://doi.org/10.1109/radar58436.2024.10993743.
Noé Lallouet, Tristan Cazenave, Cyrille Enderli and Stéphanie Gourdin. “Loss Function Design For Training Robust Radar Detectors Using Deep Learning.” Conference on Artificial Intelligence for Defense, DGA Maîtrise de l’Information, https://hal.science/hal-04328554v1.

International football machine learning challenge on synthesis of novel viewpoints for football matches.
4th place out of 65 teams and 95 submissions.

Winner of a global three-day hackathon on the subject of weather prediction in Formula One with a prize pool of 60k€+.
Placed first out of 140+ teams.

Award received following the presentation of a sports‑tech startup implementing deep learning solutions in the field of indoor boulder climbing.

Languages:
- English: Bilingual (C1), TOEIC Listening and Reading: 990/990
- French: Native language
- Spanish: Professional capability (B2)
- Japanese: Notions (A1)
Programming Languages & Frameworks: Python, NumPy, PyTorch, TensorFlow, SQL, Java, JavaScript
Scientific domains: Artificial intelligence, machine learning, optimization, probability, graphs, signal processing
Technologies: GitHub, AWS SageMaker, EC2, S3, Lambda, EFS, ECR