Mr. Angelos Athanasiadis | Neural Networks | Research Excellence
Aristotle University of Thessaloniki | Greece
Angelos Athanasiadis is a Ph.D. candidate in Electrical and Computer Engineering at Aristotle University of Thessaloniki (AUTH), specializing in FPGA-based acceleration of Convolutional Neural Networks. With expertise spanning embedded system development, heterogeneous computing, and cyber-physical systems, he has contributed to both academic and industrial innovation through participation in EU research initiatives—including the ADVISER and REDESIGN projects—and through consultancy and R&D roles at EXAPSYS and SEEMS PC. His work focuses on advancing energy-efficient hardware acceleration, leading to the development of a parameterizable HLS matrix multiplication library for AMD FPGAs that enables full-precision CNN inference for accuracy-critical domains such as aerial monitoring and autonomous embedded systems. He further expanded the field with FUSION, an open-source high-fidelity distributed emulation framework integrating QEMU with OMNeT++ via HLA/CERTI synchronization to support deterministic, timing-aware multi-node execution and realistic prototyping of heterogeneous systems. Complementing his strong technical background, he holds an MBA awarded with high distinction and an M.Eng. in electronics and computer systems, supported by internships at Cadence Design Systems in Munich.
Profiles: Orcid | Google Scholar
Featured Publications
"An efficient open-source design and implementation framework for non-quantized CNNs on FPGAs", A Athanasiadis, N Tampouratzis, I Papaefstathiou, Integration, 2025.
"Energy-Efficient FPGA Framework for Non-Quantized Convolutional Neural Networks", A Athanasiadis, N Tampouratzis, I Papaefstathiou, arXiv preprint arXiv:2510.13362, 2024.
"An Open-source HLS Fully Parameterizable Matrix Multiplication Library for AMD FPGAs", A Athanasiadis, N Tampouratzis, I Papaefstathiou, WiPiEC Journal-Works in Progress in Embedded Computing Journal 10 (2), 2024.