About me
I am a Postdoc Researcher at TU Wien, Austria. Before that, I was a postdoc at the University of Salerno, Italy, where I also did my PhD.
My research is in High-Performance Computing, with a focus on communication in large-scale HPC systems. During my PhD, I worked on optimizing MPI-level communication, looking at collective algorithms, network behavior, and how message passing can be made more efficient on modern interconnects and network topologies.
At TU Wien, my focus has shifted toward distributed deep learning. I work on GPU-GPU communication, profiling and tuning libraries like NCCL, and understanding where communication becomes a bottleneck in multi-GPU training.
I also have experience in energy-efficient HPC, including techniques such as frequency scaling to reduce power consumption in modern GPUs.
More broadly, I am interested in how HPC techniques can be applied to make large-scale AI training faster and more energy-efficient, and in parallelizing and accelerating scientific code from other domains.
.
Research Interests
High Performance Computing
Accelerating compute-intensive workloads, such as Scientific workloads, through parallelization strategies targeting multi-core, multi-node, and GPU environments.
HPC for AI
Optimizing multi-GPU deep learning training by analyzing and profiling computation and communication phases, with a focus on tuning the NCCL communication layer to improve performance.
HPC Interconnects & Topologies
Profiling and optimizing high-speed interconnects (like InfiniBand and NVLink) and large-scale network topologies (such as Dragonfly+). Focused on tuning message-passing libraries (MPI, NCCL) to minimize latency and communication overhead.
Energy-Efficient HPC
Applying power-aware practices, such as frequency scaling, to optimize the performance-to-power ratio in HPC applications.
Performance Analysis & Profiling
Characterizing and analyzing the performance of parallel and distributed workloads to identify computational bottlenecks, understand system variability, and guide optimization strategies.
Scientific Computing
Developing, parallelizing, and accelerating applications for other fields of science to bring higher HPC performance to applied domains.
News!
- Jun 2025: I presented our paper in the ASHES Workshop at IPDPS 2025, in Milan.
- Feb 2025: Our paper about GPU Frequency Scaling is accepted in the main track of IPDPS 2025!
- Oct 2024: I joined the Parallel Computing research unit at Vienna University of Technology (TU Wien) as a Postdoc.
- Sep 2024: I attended CLUSTER 2024 in Kobe, Japan, to present our paper.
- Jul 2024: Our paper is accepted in CLUSTER 2024!
- May 2024: I am now a PostDoc researcher in HPC at the University of Salerno.
- May 2024: I successfully defended my PhD thesis with an “Excellent” evaluation!
- Sep 2023: I attended EURO-PAR 2023 in Limassol, Cyprus, to present our paper in the PhD Symposium.
- Sep 2023: I’m visiting Parallel Computing group at TU Wien (Austria).
- May 2023: I attended CCGRID 2023 in Bangalore, India, to present our paper “EMPI”.
- Nov 2022: Our paper won the Best Paper Award in Bench 2022.
- Sep 2022: I attended IEEE Cluster 2022 in Heidelberg, Germany.
- Nov 2020: I started my PhD at the University of Salerno.
- Feb 2020: I defended my Master’s thesis.