Speakers
Meet the brilliant minds who will be sharing their expertise and insights at Py4AI.
See you in Pavia on May 10, 2025.
Alfredo Canziani
Assistant Professor of Computer Science
@NYU
Topic: From Classification to Conversation - The evolution of Neural Networks to Transformers
Alfredo Canziani's research focuses on machine learning for autonomous driving, particularly in uncertainty estimation, failure detection, and long-term planning. He holds degrees in Electrical Engineering from Trieste University, Cranfield University, and a PhD from Purdue University. Outside academia, he is passionate about music, dance, and cooking, and continues to expand his free online Deep Learning and PyTorch course.
Alice Casiraghi
Circular Economy and Systemic Design Expert
Topic: Tigris, Euphrates, AI
​
"The services and products we use everyday have a built-in climate impact, that originates in the way we design them. Tackling the climate impacts of everything at the source has become the main challenge for our economies to future-proof them against the scarcity and abundance of natural resources and the recent unpredictability of global supply chains. AI is no different, from the use of water, to the impact of its infrastructure, from satellites to supercomputers and data centers, to the different aims AI is used for."
Valeria Zuccoli
AI Scientist
@Bettini
Topic: Repetita Non Iuvant: Why Generative AI Models Cannot Feed Themselves
​
"As AI floods the digital landscape with content, what happens when it starts repeating itself? We will explore model collapse in image generation, revealing how previous outputs can contaminate future training and compromise the creation of diverse imagery. We will also examine the issues that lead to this degradation, and highlight the critical need to protect the evolution of visual AI models."
Daniele Zonca
Senior Principal Software Engineer
@Red Hat
Topic: Serving GenAI on Kubernetes​
"LLMs are increasing rapidly in size and capability, driving the need for scalable deployment solutions that effectively harness GPU (or in general AI Accelerator) resources across multi-node environments. In Red Hat we integrate multiple projects, like vLLM, KServe and Kubernetes, to enable scalable and flexible model-serving solutions across both uniform and mixed GPU setups. I will discuss real-world case studies to showcase how to overcome the challenges of LLM deployment in diverse environments."
Dana Aubakirova
Machine Learning Research Engineer @Hugging Face
Topic: Democratising AI for Robotics: How To Build Community-Driven Foundational Models for Robotics
"Generative AI is transforming robotics by making it more accessible, adaptable, and capable of human-like interactions. In this talk, I’ll explore how AI-driven models, such as vision-language-action models, enable robots to perform complex tasks with affordable hardware and how open collaboration through LeRobot at Hugging Face is accelerating innovation in the field."
.png)
Merve Noyan
ML Advocate Engineer
@Hugging Face
Topic: Multimodal AI
​
"Multimodal models (vision LMs) have become increasingly more capable and mainstream. In this talk, I will talk about where it all began, latest multimodality trends (multimodal safety, multimodal RAG, agents), latest sota open models and tools for you to get started in building multimodal applications."
.png)
Luca Arrotta
Lead AI Scientist
@Datapizza
Topic: AI agents architecture and best practices
​
"AI agents are redefining the way we automate business workflows, unlocking immense advantages in speed, productivity, and efficiency. However, the design of autonomous AI systems brings new challenges, from data quality and security to the reliability of modern Generative AI models. What are the patterns and best practices? How can you build modern, robust, and reliable AI agents?"
_edited.jpg)
Mirco Planamente
AI Researcher
@ARGO Vision
Topic: Egocentric Vision - AI through the Eyes of the User
​
"By moving the focus from outside observation to first-hand experiences, Egocentric Vision - AI Through the Eyes of the User - is revolutionizing AI and opening the door for hands-free technologies. This talk will outline its evolution, offering a clear snapshot of the current state of the art, the key technical challenges, and the most promising solutions. We explore how this field bridges multiple areas - from robotics to natural language processing - and its practical applications. Finally, we’ll look ahead at future directions, highlighting its potential to drive innovation across diverse sectors."
.png)
.png)
Fanny Bouton
Quantum Lead & Product Manager @OVHcloud
Topic: Why Quantum Computing when you have AI?
Fanny Bouton is the Quantum Lead & Product Manager at OVHcloud, co-founder of France Quantum, and host of Quantum Podcasts. With over 20 years of experience in emerging technologies, she is a leading voice in innovation and the future of computing. Passionate about innovation, Fanny launched her blog and the "Fanny's Party" events in the early 2000s, bringing together tech enthusiasts and demystifying complex technological subjects.
.png)
Florian Hönicke
Principal AI Engineer
@Jina AI
Topic: How to build an Agentic Search Flow
Florian has worked as a Data, Solution, and Machine Learning Engineer at companies such as SoundCloud, Deloitte Deutschland, and Visual Meta GmbH.
In his talk on Agentic Search, he explores how search has evolved from keyword-based search systems to semantic search - and why the next major breakthroughs won't come from small tweaks to the search engines, but will come from the agentic use of such systems.
.png)
Aladdin Persson
ML Engineer
@NaNEye
Topic: How LLMs are reshaping recommendation systems
​
With academic experience from Stanford University and Chalmers University of Technology, he explores how large language models (LLMs) are transforming recommender systems. His talk, “How do LLMs change the recommender system landscape?” will demonstrate how they can be used to augment and improve traditional recommender systems.
Enrico Bertino
Co-founder, Chief AI Officer
@indigo.ai
Topic: From Wow to Work: taking LLMs research into production
Deploying LLMs in real-world applications isn’t just about scale, it’s about efficiency and control. Enrico's talk explores token-efficient strategies to bridge the gap between research and industry, optimizing costs, latency, and reliability. From structured agent pipelines to model fine-tuning and open-source scalability, learn how to make GenAI truly production-ready.
.png)
Gabriele Venturi
Founder, CEO
@PandasAI
Topic: Building an AI Company (and staying sane)
Gabriele is a software engineer and serial entrepreneur with experience working in startups across Europe. Having successfully built and sold multiple projects, he is always exploring new ideas and challenges in technology, innovation, marketing, and finance. Passionate about continuous learning, he brings a multidisciplinary approach to business and product development.
Raffaele Mauro
Partner
@Primo Space
Topic: Space Economy: Technology, Markets and AI Application
​
Raffaele Mauro is a Partner at Primo Space, a venture capital fund focused on the space economy. He previously led Endeavor Italia and worked with United Ventures, P101, and Intesa Sanpaolo to support high-tech enterprises. Holding degrees from Harvard, Bocconi, and Singularity University, he is an author and a member of multiple global innovation networks. His talk will provide valuable insights into venture capital, frontier technologies, and the future of space entrepreneurship.
.png)