More About Me

Career Path

I’m now a Principal LLM Performance Architect at Volantis Semiconductor. Volantis builds photonic interconnects inside AI accelerators to extend memory reach beyond electrical links, enabling much larger, high-bandwidth memory pools for low-latency inference of frontier-scale models.

Previously, I was on the architecture team at d-Matrix, working on distributed LLM inference across mixture-of-experts (MoE), performance modeling, 3D-stacked DRAM, collectives, and scale-up/scale-out networks. My work focused on end-to-end performance analysis and memory-centric system design for high-throughput, low-latency inference. I produced the performance estimates presented in our ISCA’26 paper, Early Silicon of Raptor: The First 3D-DRAM Accelerator for Generative Inference.

Previously, I was at Rain AI for two years. As one of the initial technical hires, I embraced a role that was as varied as it was fulfilling. My day-to-day involved realizing our AI accelerator architecture—breaking the memory wall with compute-in-memory, driving radical hardware–software codesign across ML workloads (from LLM attention to CNN convolutions), and leading technical presentations and customer interactions. Patents from Rain’s compute-in-memory work were later acquired by OpenAI.

Before Rain AI, I worked as a machine learning researcher at SK hynix, focusing on the intersection of hardware, software, and CMOS image sensors (CIS) to enable efficient execution of cutting-edge computer vision workloads. Later, I co-designed compression accelerators (LZ4 and ZSTD) for computational SSDs and explored processing-in-memory (PIM) concepts on Hybrid Memory Cubes (HBMs).

In 2018, I had the enriching experience of interning on Google’s Video Understanding Team, where I contributed to designing DNN models for YouTube video analysis—an opportunity that deepened my machine learning and video-analytics expertise in a high-impact setting.

Academic Background

I got my Ph.D. and Masters’s degrees in computer science from the Georgia Institute of Technology. I had the pleasure to work with Professor Hyesoon Kim at HPArch lab. My research interests lie within computer architecture, but I am always eager to explore new areas. Specifically, I have worked on processing in and near memory, edge devices, robotics, and deep learning. My thesis focused on enabling edge and IoT devices (i.e., resource-constrained devices, robots, and drones) to execute deep neural networks (DNNs) collaboratively, efficiently, securely, and reliably. I am excited to continue this work in industry and explore new ways of pushing the boundaries of edge systems.

My academic journey began at the Sharif University of Technology, where I earned a bachelor’s degree in Electrical Engineering. In my senior year, I took on a significant role, managing our Microprocessor Systems Laboratory (MicroLab). This lab was a pivotal resource for undergraduates, being the only place offering access to advanced FPGA boards, MCUs, and processors beyond the basic 8051 model. Working with a range of technologies, from Altera to Xilinx FPGA boards, alongside robotics, microprocessors, and microcontrollers, was a defining experience. It deeply influenced my career path, steering me towards a focus in computer architecture.

Early Inspirations and Interests

Astronomy captivated me during my high school years, leading to significant achievements in national and international olympiads, including two silver and one gold medal at the IOAA. Unlike other disciplines such as mathematics or chemistry, astronomy was not a standard part of high school curriculums. This gap compelled me to embark on a self-learning journey, where I had to seek out resources independently, establish a community, and teach myself. This early experience in self-education not only fueled my passion for astronomy but also instilled in me a lifelong habit of self-directed learning. This skill has been invaluable, enabling me to seamlessly transition into electronics and computer science, and continues to be a cornerstone of my personal and professional development.

Personal Pursuits

In my leisure time, I spend time on board games or grand strategy games by Paradox Interactive to hands-on maintenance of my car, I find joy in a variety of activities. This includes figuring out and teaching myself new skills, like designing custom boards with sensors, fixing an oil-fired boiler, and wood working. These DIY projects not only satisfy my curiosity but also resonate with my advocacy for open-source software, free internet, and the right to repair. I aspire to play post-rock music. Nature also offers me solace, and I pursue outdoor sports like squash, tennis, ping pong, soccer, rock climbing (especially bouldering), cycling, skiing, and backpacking.