Mahdi Nikdast
Papers
2
Total Citations
14
H-Index
2
About
Mahdi Nikdast is a leading researcher at the intersection of silicon photonics and machine learning hardware, with a primary focus on developing novel accelerators for graph neural networks (GNNs). His most cited work, "GHOST: A Graph Neural Network Accelerator using Silicon Photonics" (2023), introduces a groundbreaking approach to overcome the fundamental memory and bandwidth bottlenecks that plague conventional electronic GNN processors. By leveraging the unique properties of silicon photonics—namely, high bandwidth, low latency, and energy-efficient data movement—Nikdast’s architecture enables dramatically faster and more efficient processing of graph-structured data. This innovation has direct implications for critical applications in recommendation systems, social network analysis, drug discovery, and robotics, where GNNs are increasingly indispensable. With over a dozen citations in a short time, GHOST is already recognized as a pivotal contribution to the emerging field of photonic computing for AI. Nikdast’s work exemplifies how cross-disciplinary thinking—merging photonic device physics with neural network design—can unlock new performance frontiers, positioning him as a key figure in the next generation of specialized, high-performance machine learning accelerators.
Research Focus
Key Achievements
Top Papers
- 1GHOST: A Graph Neural Network Accelerator using Silicon Photonics12 citations · 2023
- 2GHOST: A Graph Neural Network Accelerator using Silicon Photonics2 citations · 2023