Michail Sapkas, PhD Physicist in Edge AI and FPGAs
PhD Researcher · University of Padova · INFN Padova
Michail Sapkas
Μιχαήλ Σάπκας
|
1journal paper (JINST 2026)
6talks and posters
9conferences, schools and workshops
4active research lines
About me
I am a PhD student in Technologies for Fundamental Research in Physics and Astrophysics at the University of Padova, working with the INFN Padova group on edge AI and hardware accelerators for particle physics. I trained as a physicist, moved into deep learning through a Master's in *Physics of Data*, and found my way into hardware through FPGAs and the AMD Versal AI Engine. My work sits at the point where physics needs decisions in nanoseconds: running recurrent neural networks fast enough for detector readout, for accelerator control, and for the CMS trigger at CERN. I am also part of the BOOST Lab group, with interests in particle physics and the CMS experiment.Right now
Research
GRU inference on the Versal AI Engine
Latency-constrained recurrent networks on the VCK190, with a hybrid AIE and FPGA dataflow.
ProjectCNN trigger for a CAEN digitizer
A real-time 1D CNN that detects particle hits directly in the digitizer FPGA.
DetectorMini Drift Tubes muon telescope
Time-to-digital pipeline and machine learning for muon tracking and timing.
Latest news
Sep 2026
Visiting researcher at SLAC National Accelerator Laboratory, Stanford, from September 2026
Apr 2026
Talk at RL4AA'26, 4th collaboration workshop on Reinforcement Learning for Autonomous Accelerators, Liverpool
2026
Paper on the latency-constrained GRU in the Versal AIE published in the Journal of Instrumentation (JINST 21, C03015)
Oct 2025
Talk at TWEPP 2025, Topical Workshop on Electronics for Particle Physics, Crete
Moments
With the Versal AI Engine VCK190 evaluation kit, in my temporary office at IPE.
Visiting Mare Nostrum 5 at the Barcelona Supercomputing Center.