I’m a fourth-year PhD student at the Scalable Parallel Computing Laboratory (SPCL) in the Department of Computer Science at ETH Zürich, advised by Prof. Torsten Hoefler. I’m also a student researcher at Microsoft, where I’ve been working with Abdul Kabbani for over three years.
My research focuses on datacenter networking, especially congestion control, load balancing, AI/HPC workloads, collective communication, and network topologies. I also contribute to the Ultra Ethernet Consortium (UEC) effort, including work on congestion-control and load-balancing proposals and related tooling (mostly HTSIM).
I received my BSc in Computer Science and Engineering from the University of Bologna and my MSc in Computer Science from ETH Zürich. During my studies, I interned at the European Space Agency (ESA) and Amazon.
Outside of research, I enjoy reading about space and space exploration, watching sci-fi movies, skiing, making (and cooking) pizza, and visiting theme parks.
Below are some of my recent publications—feel free to reach out if you'd like to discuss any of them!
2026
-
PICO: Performance Insights for Collective Operations
Saverio Pasqualoni, Tommaso Bonato, Lorenzo Piarulli, and 3 more authors
In ISC High Performance 2026 Research Paper Proceedings (41st International Conference), Hamburg, Germany, Jun 2026
ISC 2026 Hans Meuer Award (Best Paper)
ISC High Performance 2026 Hans Meuer Award for the best research paper of the year (announcement).
Collective operations are cornerstones of both HPC applications and large-scale AI training and inference, yet benchmarking them in a systematic and reproducible way remains difficult on modern systems due to the complexity of their hardware and software stacks. We present PICO, an open-source framework that decouples portable experiment setup from platform execution, provides a backend-adaptive parameter selection interface across MPI and NCCL, supplies plain-MPI reference collective implementations, and records system configuration for reproducible comparisons. Evaluated on three major supercomputers, PICO shows that default collective algorithms and transport settings can be up to 5x slower than the best available choice. Benchmark-informed tuning with PICO reduces replayed LLM training times by up to 44%.
-
Flowcut Switching: High-Performance Adaptive Routing with In-Order Delivery Guarantees
Tommaso Bonato, Daniele De Sensi, Salvatore Di Girolamo, and 4 more authors
IEEE/ACM Transactions on Networking, Jan 2026
-
REPS: Recycled Entropy Packet Spraying for Adaptive Load Balancing and Failure Mitigation
Tommaso Bonato, Abdul Kabbani, Ahmad Ghalayini, and 7 more authors
In Proceedings of the European Conference on Computer Systems (EuroSys’26), Apr 2026
2025
-
ATLAHS: An Application-centric Network Simulator Toolchain for AI, HPC, and Distributed Storage
Siyuan Shen*, Tommaso Bonato*, Zhiyi Hu, and 3 more authors
In Proceedings of the International Conference for High Performance Computing, Networking, Storage and Analysis (SC’25), Nov 2025
Best Student Paper Finalist
-
Uno: A One-Stop Solution for Inter- and Intra-Datacenter Congestion Control and Reliable Connectivity
Tommaso Bonato*, Sepehr Abdous*, Abdul Kabbani, and 11 more authors
In Proceedings of the International Conference for High Performance Computing, Networking, Storage and Analysis (SC’25), Nov 2025
2024
-
Swing: Short-cutting Rings for Higher Bandwidth Allreduce
Daniele De Sensi, Tommaso Bonato, David Saam, and 1 more author
In 21st USENIX Symposium on Networked Systems Design and Implementation (NSDI ’24), Apr 2024
2022
-
HammingMesh: A Network Topology for Large-Scale Deep Learning
Torsten Hoefler, Tommaso Bonato, Daniele De Sensi, and 7 more authors
In Proceedings of the International Conference for High Performance Computing, Networking, Storage and Analysis (SC’22), Nov 2022
SC22 Reproducibility Advancement Award and invited as CACM Research Highlight
