CV
Théo Taburet
Data Scientist at Ezako · Antibes, France
theotaburet.github.io · theo.taburet@protonmail.com · github.com/theotaburet · linkedin.com/in/theotaburet · orcid.org/0000-0001-8165-6826
Where I have worked, what on, and what came out of it.
Experience
Data Scientist
Ezako, Antibes, France · current position
- Object detection (YOLO, RF-DETR) to annotate underwater sound
- Anomaly detection on MEMS sensors with quantised 1D CNNs running on STM32 microcontrollers
Postdoctoral Researcher in Computer Vision and Image Processing
Université de La Rochelle, L3i laboratory / Yooz · 2021-2023
Deep learning to detect forged business documents, with Yooz.
- Double JPEG compression detection with Siamese CNNs
- Steganalysis to locate manipulations in an image
- A receipt dataset for document forgery detection, published at ICDAR 2023
- FastAPI services to integrate the models
Teaching Assistant
Université de La Rochelle & Excelia · 2022-2023
- Web technologies (L1 & L2)
- Introduction to Computer Vision (M2)
R&D Mechanical Engineering Apprentice
Synerlink S.A. · 2014-2017
Designed and sized thermoforming moulds, and tuned the process for production. Took part in managing R&D projects.
Education
PhD, Applied Mathematics (Steganography)
École Centrale de Lille, CRIStAL laboratory, Lille, France · 2017-2020
Thesis: Méthodes de stéganographie fondées sur la prise en compte du bruit de capteur (steganography methods based on sensor noise), defended in October 2020.
MSc, Digital Signal and Image Processing
Cranfield University, Cranfield, UK · 2016-2017
Engineering Degree, General Engineering
ESTIA, Bidart, France · 2014-2017
Multidisciplinary engineering: mechanical, electronics, and computer science.
Skills
| Area | Skills |
|---|---|
| Machine Learning | Deep Learning · CNNs · Object Detection (YOLO, RF-DETR) · Anomaly Detection · Segmentation · Fraud Detection |
| Embedded ML | 1D CNNs · Quantisation · STM32 Microcontrollers · MEMS Sensors |
| Signals and Images | Signal Processing · Image Processing · Steganography · Steganalysis · Image Forensics · Object Tracking |
| Languages | Python · C/C++ · HTML/CSS/JS · SQL · MATLAB · Bash |
| Frameworks | PyTorch · TensorFlow · Keras · OpenCV · FastAPI · scikit-learn |
| Tooling | Git · Docker · CI/CD · GitHub Actions · GitLab CI/CD |
Projects
diapason
Since 2026 · diapason.fm
One person picks the music, everyone nearby scans a QR code and plays it out loud on their own portable speaker, in sync. Runs in the browser, nothing to install. Rust server, WebAssembly client, in production.
Ravitools
GPX enrichment with offline points of interest for long-distance cyclists.
Volunteering
Co-founder and organiser, Pignon des Charentes
La Rochelle, France · since 2023
Bike treasure hunts built around riddles, leading participants through the city to a final gathering point. I create the visual content, design the riddles, and host the events.
Languages
French (native) · English (fluent)
Certificates
- Python, France Université Numérique, 2021
- Scikit-learn, France Université Numérique, 2021