Internship, Ph.D., Postdocs, and Research Engineer positions

Feel free to contact me (paul.boniol@inria.fr) if you are interested in one of the positions below. Unfortunately, I am receiving a large volume of emails, and I struggle to respond in time. I apologize in advance.

🔬 Postdoc Position

Role: Postdoctoral Research Intern – Bridging knowledge and temporal data

Location: École Normale Supérieure (ENS) and EDF lab (Chatou)
Host Institution: INRIA Paris or EDF
Duration: 1 or 2 years
Project: Bridging knowledge and temporal data

Position Overview

I am looking for a postdoctoral student working on temporal data analysis and on temporal knowledge representation. More details on the topic soon. If you graduate around May 2027 and you are interested in conducting research with an industrial partner (EDF), please contact me (paul.boniol@inria.fr).

🔬 Research Internship Position

Role: Master’s Research Intern – Evaluating Large Language Models on Time Series Analysis Tasks

Location: École Normale Supérieure (ENS), Paris, France
Host Institution: INRIA Paris & ENS-PSL (VALDA Team)
Duration: 6 months (Flexible)
Project: Evaluating Large Language Models on Time Series Analysis Tasks

Position Overview

This project aims to systematically evaluate the capabilities of several frontier LLMs, including reasoning-enabled models, across three fundamental time series mining tasks: anomaly detection, motif discovery, and segmentation. The performance of LLMs will be compared against both state-of-the-art algorithms and human analysts, across diverse datasets and task-specific questions. Please send a detailed CV and a brief Statement of Interest to:

Please send a detailed CV and a brief Statement of Interest to:

Roberto Stanzione: roberto.stanzione@inria.fr
Paul Boniol: paul.boniol@inria.fr

🔬 Research Internship Position

Role: Master’s Research Intern – Differentiable solvers for time series segmentation

Location: Université Rennes or École Normale Supérieure (ENS), Paris, France
Host Institution: Université Rennes or Inria Paris & ENS-PSL (VALDA Team)
Duration: 6 months (Flexible)
Project: Differentiable solvers for time series segmentation

Position Overview

Time series segmentation (change-point detection, piecewise approximation, regime discovery) relies on discrete optimization that is non-differentiable, blocking its use as a layer in trainable pipelines. Smoothing a fixed-grid dynamic program is well understood; the real obstacle is model selection, since the optimizer also chooses how many segments to use. This internship asks whether that segment-count decision can itself be made differentiable, end-to-end.

Please send a detailed CV and a brief Statement of Interest to:

Romain Tavenard: romain.tavenard@univ-rennes2.fr
Paul Boniol: paul.boniol@inria.fr