Offre en lien avec l’Action/le Réseau : – — –/– — –

Laboratoire/Entreprise : Laboratoire DVRC
Durée : 4-6 mois
Contact : christophe.rodrigues@devinci.fr
Date limite de publication : 2026-02-02

Contexte :

Sujet :
M2 Research Internship

Generative AI for Ransomware Time‑Series Detection

The context of the work is Owlyshield, a behavioral time‑series dataset built from Endpoint Detection and Response (EDR) logs to detect ransomware attacks. We aim to generate realistic synthetic sequences that preserve temporal structure and inter‑feature dependencies to improve anomaly detection (benign or ransomware) on this dataset.
Missions
The intern will:

review deep generative models for time series (with a focus on transformer‑style methods);
design and implement conditional generative models (e.g., TimeGAN‑like architectures) to produce labeled multivariate sequences consistent with Owlyshield statistics;
define quality metrics (distributional similarity, temporal coherence, correlation structures);
evaluate the impact of synthetic data on downstream ransomware/anomaly detection performance.

Profile
M2 student in Computer Science, Data/AI, or Applied Mathematics. Solid background in machine learning and Python; experience with deep learning (PyTorch or TensorFlow) and time‑series data. Knowledge of generative models (GANs/VAEs) and transformers, cybersecurity is a plus. Autonomy, rigor, and good written English are expected.
Practical information
Duration: 4-6 months (full-time, 2026).
Location: ESILV, Paris
Supervision: Sourav Rai, Christophe Rodrigues and Nga Nguyen
Application
Send CV, cover letter, grades and recommendation letters to christophe.rodrigues@devinci.fr and nga.nguyen@devinci.fr with subject: “M2 Internship – Generative Time‑Series”.
Dataset/ References
[1] SitInCloud, “Ransomware Detection (Owlyshield documentation),” https://docs.sitincloud.com/concepts/ransomware-detection.html
[2] J. Yoon, D. Jarrett, M. van der Schaar, “Time‑series Generative Adversarial Networks,” NeurIPS, 2019.
[3] X. Li, V. Metsis, H. Wang, A. H. H. Ngu, “TTS-GAN: A Transformer-based Time-Series Generative Adversarial Network,” in Proc. 20th Int. Conf. on Artificial Intelligence in Medicine (AIME), 2022.

Profil du candidat :

Formation et compétences requises :

Adresse d’emploi :
Paris, La Défense