About Me
I am a student in Computer Engineering at the University of California, San Diego, advised by Prof. Hadi Esmaeilzadeh. My research interests center on computer architecture, with a particular focus on designing custom hardware accelerators for robotics and secure LLM inference, as well as repurposing old hardware to support sustainable, green computing.
News
- Oct 2024 Presented our research paper "Reducing the Carbon Footprint of EdTech with Repurposed Devices" at the IEEE 15th International Green and Sustainable Computing Conference (IGSC) in Chicago!
- Jun 2024 Joined the ACT Lab at UC San Diego as a Researcher under the guidance of Prof. Hadi Esmaeilzadeh.
- Jun 2023 Began research internship at the Qualcomm Institute working on head scan acoustics systems.
Publications
"Reducing the Carbon Footprint of EdTech with Repurposed Devices"
IEEE 15th International Green and Sustainable Computing Conference (IGSC) • Oct 22, 2024
@inproceedings{switzer2024reducing,
title={Reducing the Carbon Footprint of EdTech with Repurposed Devices},
author={Switzer, John and Katel, Suman and Lee, Jun Chul and Rajan, Ashwin Rohit Alagiri and Kastner, Ryan and Pannuto, Pat},
booktitle={IEEE 15th International Green and Sustainable Computing Conference (IGSC)},
pages={1--8},
year={2024},
organization={IEEE}
}
Experience
Researcher
Worked under Professor Hadi Esmaeilzadeh at the ACT Lab for robotics accelerator and Secure LLM Inference accelerator projects. Developed analytic and cycle-level simulator for systolic array based robotics workload accelerator. Architected visualization tools and GPU benchmarking scripts.
Undergraduate Researcher & Teaching Assistant
Co-authored and published research on GPU workload performance on repurposed Android devices to reduce the carbon footprint of EdTech (published at IGSC/MICRO 2024). Modernized the CSE 160 (Parallel Programming) curriculum from CUDA to OpenCL. Served as Teaching Assistant in Winter 2025.
Intern
Developed systems for computing Head-Related Transfer Function (HRTF) models based on 3D Head-and-Torso scans. Tested models in AR/VR setups, and initiated moving the audio raytracing system from CUDA to Apple Metal for better cross-platform compatibility.
Selected Projects
2D Systolic Array Accelerator
Architected a 2-bit/4-bit reconfigurable systolic array with a reconfigurable weight and output stationary setup. Trained and implemented 2-bit and 4-bit quantized (QAT) VGG-16 and BERT models for machine learning workloads.
Autonomous UAV
Developed an autonomous unmanned aerial vehicle (UAV) utilising PX4 flight stacks and NVIDIA Jetson Nano platforms. Implemented real-time computer vision algorithms for obstacle avoidance, visual odometry, and SLAM.
Relativity Simulator
A high-performance C++ based physical renderer for visualizing special relativistic physics phenomena (like Lorentz contraction, Doppler shifts, and light aberration) using ray-tracing techniques.