About Me

Hi! I am Paolo Notaro :wave:,
I am a Machine Learning Engineer from Italy, based in Munich, Germany 🇮🇹 🤌🏼 🇩🇪 . On this website I enjoy discussing of machine learning, IoT, and linguistics.

In my day-to-day work, I tinker with Large Language Models and NLP, build ML workflows through MLOps and DevOps, and explore confidential computing and model evaluation. From 2020 to 2024, I was a PhD Student at the Technical University of Munich (TUM). My research focused on the application of AI to manage and operate large-scale computing systems, a research area known as AIOps. As part of my PhD, I joined Huawei's Intelligent Cloud Technlogies Lab in Jorge Cardoso's team, head of the Ultra-scale AIOps research team.

Previously, I earned a BSc in Computer Engineering from the Polytechnic of Turin (PoliTO), Italy in 2017 and a MSc in Informatics from the Technical University of Munich (TUM), Germany in 2019.

In the past, I was involved in several other projects related to machine learning and security. In 2018, I interned at Crashtest Security to improve efficiency and precision of their vulnerability testing service. In 2018-2019, I was Teaching Assistant for the "Intro to Deep Learning" class at TUM. In 2019, I wrote my Master Thesis at Airbus Defence and Space on Radar Emitter Classification using RNNs.

You can find my resume here .

ML Engineer, IABG

October 2023 — present

  • Lifecycle management of LLM-based services, including RAG, agents, and chatbots, along with research into new AI-driven product concepts.
  • Work on confidential computing and privacy-enhancing techniques (DP, HE, TEE, anonymity), as well as assessing ML privacy risks such as MIA, model extraction, and model inversion.
  • Hosted two editions of the AI Microclass at TUM Venture Labs.
  • Extensive experience in AI evaluation and benchmarking to assess performance, robustness, and reliability.
  • Some exploration of Reinforcement Learning applications in cybersecurity.

    • PhD Student @ TUM + Huawei

    January 2020 — September 2023

    AIOps for Online Failure Prediction and Root-cause Analysis (RCA)

  • I developed and deployed in production an NLP-based solution for CLI security, leading to the discovery of 10+ security concerns and the publication of a patent as main inventor.
  • I developed a new Root-cause Analysis system, resulting in 37x faster execution with increased explanation accuracy (+15%).
  • I developed novel forecasting and neural-based classification methods for Hardware Failure Prediction, improving prediction accuracy (F1 +13%)
  • I published as main author the very first survey on AIOps, providing a comprehensive overview of topics, methods, tools, data sources and target systems. In total, I was the main author of 5+ publications in high-rank conferences and journals (A/A*) and I supervised 10+ students in university projects, including seminars, master theses and guided research

    • MSc Student, Informatics @ TUM

    October 2017 — November 2019

  • Main Topics: AI, Machine Learning, Deep Learning, Computer Vision, Natural Language Processing
  • Master Thesis: Classification of Electromagnetic Pulse Signals with Deep Learning
  • Final grade: 1.7

    • BSc Student, Computer Engineering @ PoliTO

    September 2014 — July 2017

  • Main Topics: Algorithms and Data Structures, Operating Systems, Digital Electronics, Control & Signal Theory
  • Grade: 108/110