Ashvin Verma is an undergraduate researcher at UC Berkeley with a decade of hands-on experience bridging software, hardware, and physics to accelerate domain-specific workloads. He contributes to Merlin, an MLIR/IREE-based compiler stack, and has prototyped lowering paths to spatial arrays like Gemmini that yield 5–30x runtime improvements through AOT optimizations. Past internships at AMD and Armory saw him build a tiling-based DSL that emits MLIR kernels and validate DSP/ML blocks on real-time RF hardware, highlighting his full-stack approach from compiler passes to embedded board support. He also teaches large-format hands-on circuits labs, pairing pedagogical clarity with deep systems expertise. Known for combining evolutionary search agents with compiler infra, he brings an experimental, results-driven mindset to ML accelerator software and hardware co-design.
10 years of coding experience
Bachelor of Science, Electrical Engineering & Computer Science, Bachelor of Science, Electrical Engineering & Computer Science at University of California, Berkeley
Contributions:20 pushes, 2 branches in 3 years 4 months
Find and Hire Top DevelopersWe’ve analyzed the programming source code of over 60 million software developers on GitHub and scored them by 50,000 skills. Sign-up on Prog,AI to search for software developers.