🎓 Education

  • Doctoral School: ED STIC – Sciences et Technologies de l’Information et de la Communication
    Université Paris-Saclay (November 2025 – Present)

    • PhD program under the CEA-LIST institute, focusing on large multimodal models and their spatial–temporal understanding.
    • Research supervised within the Artificial Intelligence and Data Science theme.

  • Master in Machine Learning for Data Science
    Université Paris Cité (September 2024 – Present)

    • Focus: Machine Learning, Deep Learning, NLP, Computer Vision, Reinforcement Learning, and Recommender Systems.

  • Engineer and Master in Computer Science (Specialization: Intelligent Systems and Data)
    École Nationale Supérieure d’Informatique (September 2019 – July 2024)

    • Focus: Probability and Statistics, Distributed AI, Stochastic Simulation, NLP, and Machine Learning.

đź’Ľ Professional Experience

  • PhD Candidate in Artificial Intelligence
    CEA-LIST, Université Paris-Saclay (March 2025 – Present)

    • Research focus: large multimodal models and their understanding of space and time.
    • Investigating how AI systems can develop richer spatial–temporal representations and reasoning capabilities.
    • Bridging multimodal perception and reasoning to enhance next-generation intelligent systems.

  • Large Language Models Research Intern
    IRT SystemX, Saclay, France (March 2025 – September 2025)

    • Adapted large language models (LLMs) for multimodal time series analysis, anomaly detection, and forecasting.
    • Developed a retrieval-augmented generation (RAG) system for intelligent diagnosis from time series data.
    • Designed a novel method for evaluating LLM outputs without relying on ground truth, leveraging semantic consistency and model-based agreement metrics.
    • Contributed to improving LLM performance and interpretability in temporal reasoning and forecasting tasks.

  • Research Intern in Machine Learning
    New York University Abu Dhabi (Remote) (August 2023 – September 2024)

    • Developed biologically inspired neural architectures for robust image classification and hashing.
    • Proposed a novel deep hashing method achieving over 90% mAP on ImageNet.
    • Co-authored publications at ECIR 2025 and arXiv.

  • AI Instructor
    Code Labs Academy (Remote) (January 2024 – September 2024)

    • Taught advanced machine learning and generative AI concepts including Transformers, RNNs, VAEs, and GANs.
    • Mentored students on applied AI projects and deep learning research topics.

  • Data Science Intern
    Bank of Algeria (August 2022 – September 2022)

    • Built inflation forecasting models using time series analysis to evaluate macroeconomic trends.

đź“š Publications

  • Sign-Symmetry Learning Rules are Robust Fine-Tuners
    arXiv preprint, 2025
    Mehdi Zakaria Adjal, Aymene Berriche, Riyadh Baghdadi
    [arXiv:2502.05925]

  • Leveraging High-Resolution Features for Improved Deep Hashing-based Image Retrieval
    European Conference on Information Retrieval (ECIR 2025)
    Aymene Berriche, Mehdi Zakaria Adjal, Riyadh Baghdadi
    [arXiv:2403.13747]

  • A Novel Hybrid Approach Combining Beam Search and DeepWalk for Community Detection in Social Networks
    WEBIST 2023
    Aymene Berriche, Marwa NaĂŻr, Kamel Yamani, Mehdi Zakaria Adjal, et al.
    [scitepress.org]