EN FR

Hugo Hadi Dayekh

Photo Hugo Dayekh
GitHub LinkedIn Cooking

Summary

Hey reader! I'm a Research Engineer at Verimag (Grenoble), working on STLRom, a Signal Temporal Logic monitoring and falsification toolbox used to analyze and guide the behavior of AI and reinforcement learning agents. I hold a PhD in computer science, where I built machine learning methods for identifying hybrid dynamical systems from time series data. Think of it as figuring out the hidden rules governing a system's behavior, and when those rules switch, just by watching it evolve. Before that, an MSc in data science and a BSc in physics and computer science. I like turning messy data into models that actually explain something, and I'm now looking to bring that mindset to industry. Outside of work, I bake, cook, and keep a running list of songs I can't stop playing.

Work & Research

Research Engineer @ Verimag, Grenoble 12/2025 – present

Working on STLRom, extending this Signal Temporal Logic (STL) monitoring tool (C++ and Python) for the analysis and guiding of AI and reinforcement learning agent behavior.

PhD Researcher @ Verimag, Grenoble 09/2021 – 05/2025

My thesis developed two machine learning methods for identifying hybrid dynamical systems, systems whose behavior switches between different modes, directly from time series data, combining passive learning (fitting a model to a fixed dataset) with active learning (querying the system to resolve where the current model is wrong). Along the way I worked on data preprocessing, incremental regression, optimization, and classification, and tested everything on both simulated and real-world experiments.
Successfully defended on 11/04/2024.
My manuscript (with a summary in French) is available here.

Visiting PhD Researcher @ Kyoto University, Japan 03/2023 – 04/2023

During my stay in Japan, I enriched my PhD work with meetings and discussions with colleagues working on similar subjects. This led to applications of my work on experimental data from Mitsubishi Electric Tokyo.

Other Work

My journey includes multiple research assistant positions and a developer position which can be further seen on my CV in English and in French.

Education

Publications

  1. N. Kochdumper, C. Hsieh, H. Dayekh, A. Gurung, T. Dang, M. Waga, H. Segawa, and K. Suenaga, “Evaluating Hybrid Automata Learning Tools Based on Their Success in Formal Verification.”
    [IEEE Xplore]
  2. H. Dayekh, N. Basset, and T. Dang, “Learning of Switched Nonlinear Dynamical Systems: Passive and Active Approaches,” in Design and Verification of Cyber-Physical Systems: From Theory to Applications, Lecture Notes in Computer Science, Springer.
    [Springer] [Book]
  3. H. Dayekh, N. Basset, and T. Dang, “Active Learning of Switched Nonlinear Dynamical Systems,” in 2024 IEEE 63rd Conference on Decision and Control (CDC), Dec. 2024.
    [IEEE Xplore] [HAL]
  4. H. Dayekh, N. Basset, and T. Dang, “Hybrid System Identification through Optimization and Active Learning,” IFAC-PapersOnLine, vol. 58, no. 11, pp. 87–92, 2024.
    [DOI] [HAL]

Awards

Extras