KshitijLakhani/VLSI_Verilog_Projects
RTL Synthesis for Fast Arithmetic circuits like Booth encoded Multipliers, Carry Save Adders, Fixed-Point and Floating-Point conversions, Control circuits like State Machines, and DSP applications like FFT.
Deep Learning Performance Engineer @ NVIDIA. MS in ECE @ UC Davis. Interested in Machine Learning and Parallel Computing.
Repositories
KshitijLakhani/TransformerEngine
A library for accelerating Transformer models on NVIDIA GPUs, including using 8-bit floating point (FP8) precision on Hopper and Ada GPUs, to provide better performance with lower memory utilization in both training and inference.
KshitijLakhani/maxtext
A simple, performant and scalable Jax LLM!
KshitijLakhani/Cache_Profiling
Matrix multiplication is used to investigate and optimize blocking factor, associativity and size of cache hierarchy. Attributes like miss rate and execution time are tabulated and compared.
KshitijLakhani/Intro_to_Parallel_Computing
Efficient Implementation of CUDA Algorithms like scan, reduce, histogram, sort and search among others via applications like blurring of images, red eye removal, searching and histogram equalization
KshitijLakhani/DNN_Machine_Learning
Implementation of some basic Machine Learning Tasks like logistic and linear regression, simple image classification, Deep CNNs, etc. using python and keras primarily
KshitijLakhani/facial-expression-recognition
Deep CNN model using data augmentation with a very high accuracy of 78.51% on Kaggle's fer2013 dataset
KshitijLakhani/Trial
Trial space