Mohak Chadha designs infrastructure for enterprise AI applications at NVIDIA. He tackles the scaling constraints of AI inference, breaking down challenges like autoregressive decoding, memory bandwidth, and scheduling under contention.
Before this, Mohak managed cloud infrastructure and security at Firebolt. His roots are in distributed systems and parallel computing, which he regularly discusses at events like KubeCon and Open Source Summit. Based in Munich, he focuses on the realities of building and shipping AI at scale.
