Gianni Pellegrini

I live in Trento, in the middle of the Italian Alps

I work as AI Scientist at Ipazia S.p.A., where I design and build cutting-edge artificial intelligence products for businesses, combining traditional machine learning with modern agentic AI systems.

My current work spans time-series analysis for behavioral profiling, recommendation systems, financial risk estimation, and the research and development of agentic AI solutions built on top of frameworks such as LangGraph and Autogen.

I hold a Ph.D. in Computer Science and Telecommunications from the University of Trento, where I worked under the supervision of Andrea Passerini on neuro-symbolic learning, statistical relational learning, and graph neural networks. I was also one of the organizers of the PyTorch Geometric Tutorial, an online series on Graph Neural Networks available on YouTube.

Outside of work, my interests include videogames, dungeons and dragons, climbing and enjoying life :)

news

May 01, 2026	Our paper SokoBench: Evaluating Long-Horizon Planning and Reasoning in Large Language Models was accepted at TMLR.
Mar 01, 2023	Started as AI Scientist at Ipazia S.p.A., working on machine learning and agentic AI products for businesses.
Aug 21, 2021	Our paper Learning Aggregation Functions was accepted at IJCAI 2021.
Feb 01, 2021	Co-launched the PyTorch Geometric Tutorial series — an online series on Graph Neural Networks available on YouTube.

latest posts

Mar 26, 2025	a post with plotly.js
Dec 04, 2024	a post with image galleries
May 14, 2024	Google Gemini updates: Flash 1.5, Gemma 2 and Project Astra

selected publications

TMLR

SokoBench: Evaluating Long-Horizon Planning and Reasoning in Large Language Models

Sebastiano Monti, Carlo Nicolini, Giovanni Pellegrini, and 2 more authors

Transactions on Machine Learning Research, 2026

Bib

@article{monti2026sokobench,
  title = {SokoBench: Evaluating Long-Horizon Planning and Reasoning in Large Language Models},
  author = {Monti, Sebastiano and Nicolini, Carlo and Pellegrini, Giovanni and Staiano, Jacopo and Lepri, Bruno},
  journal = {Transactions on Machine Learning Research},
  issn = {2835-8856},
  year = {2026},
  url = {https://openreview.net/forum?id=pLosAkOoGU},
  note = {},
}

IJCAI

Learning Aggregation Functions

Giovanni Pellegrini, Alessandro Tibo, Paolo Frasconi, and 2 more authors

In Proceedings of the Thirtieth International Joint Conference on Artificial Intelligence, IJCAI-21, 2021

Bib

@inproceedings{pellegrini2021learning,
  title = {Learning Aggregation Functions},
  author = {Pellegrini, Giovanni and Tibo, Alessandro and Frasconi, Paolo and Passerini, Andrea and Jaeger, Manfred},
  booktitle = {Proceedings of the Thirtieth International Joint Conference on Artificial Intelligence, {IJCAI-21}},
  year = {2021},
  url = {https://www.ijcai.org/proceedings/2021/0356},
}

RecSys

No more ready-made deals: constructive recommendation for telco service bundling

Paolo Dragone, Giovanni Pellegrini, Michele Vescovi, and 2 more authors

In Proceedings of the 12th ACM Conference on Recommender Systems, 2018

Bib

@inproceedings{dragone2018no,
  title = {No more ready-made deals: constructive recommendation for telco service bundling},
  author = {Dragone, Paolo and Pellegrini, Giovanni and Vescovi, Michele and Tentori, Katya and Passerini, Andrea},
  booktitle = {Proceedings of the 12th ACM Conference on Recommender Systems},
  pages = {163--171},
  year = {2018},
  publisher = {ACM},
}