Mani Ananth

Mani Ananth

Computer Architect & Software Engineer

I'm a Computer Architect and Software Engineer passionate about building high-performance systems. I enjoy working at the intersection of hardware and software, exploring topics like HW/SW codesign for ML applications, compilers, and kernel programming.

Work Experience

  • I am currently with Google DeepMind optimizing the Gemini family of models for inference on various platforms like TPUs and GPUs. Prior to GDM, I was working with Cerebras Systems and NVIDIA.
  • At Cerebras, I was a performance lead for their wafer-scale engine based AI systems, optimizing training workloads.
  • At NVIDIA, I worked on high performance kernel development for Deep Learning and HPC applications, and contributed to the CUTLASS open source library. Before that, I worked on the microarchitecture of Volta, Turing, Ampere, and Hopper SM including tensor cores.