Jashojit Mukherjee — AI Researcher

At NVIDIA, I work on making Deep Learning workloads run faster on our platforms. This includes optimizing and speeding up things at multiple levels including AI frameworks ( like Pytorch, Tensorflow, Jax, etc) , Model deployment for inferencing (TensorRT and Triton) , optimizing for different NVIDIA HW and architectures. I work on profiling workloads end to end, finding gaps and optimization opportunities in the workflow. I also write custom CUDA plugins for fusing operations and layers of complex DNNs. Currently I work on diffusion based models for Generative AI and help in optimization of NVIDIA Cosmos models. In the past I have worked on Quantization techniques for computer vision DNNs , AI for finance, profiling and benchmarking of Large Language Models ( LLMs) on hyperscale Multi Node Multi GPU clusters, explainability of AI models using Shapley Values. I also have experience in monocular depth prediction, image enhancement and other computer vision problems. Before joining my current role I also worked on Tegra SOC based embedded systems. I helped develop drivers for NVIDIAs automotive platform and also on ADAS Vision problem statements. I have a MTech degree from IIIT Hyderabad, where most of my courses and electives were focused on ML, AI and computer vision.

Stackforce AI infers this person is a Deep Learning and AI Optimization expert in the tech industry.

Location: Bengaluru, Karnataka, India

Experience: 10 yrs 7 mos