Girish has spent almost 4 years at Cruise, and has played an integral role in their rapid advances of the past couple of years, seeing them lead the way on the journey to full autonomy. He works in core technology areas, developing compilers and tools for easy implementation and deployment of high-performance computer vision and deep learning applications on GPUs. With his foundations in program analysis, computer architecture and compiler design, Girish likes the challenge of marrying the needs of high-end AI applications with low-level hardware constraints imposed by GPUs and other accelerators. He is now Senior Director of MLA at Cruise, which is responsible for delivering real-time on-road performance for AV software. His team achieves this by building tools (AI deployment tools, DL optimization tools and system performance diagnosis/analysis/testing tools) as well as by optimizing the software to leverage various forms of parallelism on different hardware accelerators. They also work closely with AV hardware teams to ensure they get that performance at the lowest possible hardware cost.