Francesco Sammarco

francescos.png

Location

: Tubingen, Germany

Francesco Sammarco is a AI/ML Engineer at Live Reply, working on Agentic AI and MLOps for telco related applications.

He was a PhD student at the University of Tübingen, affiliated with IMPRS, working with the Mental Health Mapping Lab on machine learning methods for complex mental health challenges. His current work explores explainable AI and generative modeling for neuroimaging, with an emphasis on understanding model reasoning and building reliable pipelines for large-scale medical data. He has co-authored work on medical image segmentation model merging (MedSAMix) and earlier published in recommender systems research, bridging applied engineering with research-driven innovation. He also worked on Reinforcement Learning for Trading, focusing on non-stationary RL for Forex trading with automatic market regime clustering. Today, he keeps working as a researcher in spare time.

Before moving into his current role and research, he worked as a DevOps engineer and telco consultant at Hewlett Packard Enterprise, where he built CI/CD automation and production-grade backend integrations, experience he now applies to reproducible ML workflows.

Selected Publications

  1. AAAI
    medsamix.png
    Medsamix: A training-free model merging approach for medical image segmentation
    Yanwu Yang, Guinan Su, Jiesi Hu, and 3 more authors
    arXiv preprint arXiv:2508.11032, 2025
    DOI PDF
  2. RecSys - ACM
    Lightweight model for session-based recommender systems with seasonality information in the fashion domain
    Nicola Della Volpe, Francesco Sammarco, Lorenzo Mainetti, and 7 more authors
    In Proceedings of the Recommender Systems Challenge 2022, 2022
    DOI PDF