I'm Harsh Gujarathi
Software Engineer and ML Systems Researcher
I'm a Graduate Student Researcher at the NEXS Lab at Georgia Tech, pursuing an MS in Computer Science. I previously worked as a Software Engineer at Microsoft on Azure Migrate.
I build backend and distributed systems that make AI products reliable in production. My recent work spans RAG agents, evaluation pipelines, event-driven services, and efficient LLM inference.
At Georgia Tech, I am studying self-speculative decoding for Mixture-of-Experts (MoE) models and profiling vLLM inference on NVIDIA H200 GPUs. I also enjoy turning real problems into useful tools, from job discovery to GRE practice.
Outside of work, I spend a good amount of time at the gym. It's my go-to for staying focused, energized, and pushing my limits both physically and mentally. I also love playing board and card games with friends—there's something thrilling about strategy and friendly competition. When I'm not lifting weights or gaming with friends, you'll probably find me exploring fictional worlds through manga or video games. I'm forever inspired by creators like Eiichiro Oda and Stan Lee for shaping these incredible universes.
Building systems, learning constantly, and open to interesting problems.