Custom PyTorch (Needle)
Implemented a subset of PyTorch from scratch that supports autograd, NDArrays, common deep-learning architectures, and acceleration using custom CUDA code.
Here is a collection of projects I've done as part of coursework or in my free time. Click on a project to see more details!
Implemented a subset of PyTorch from scratch that supports autograd, NDArrays, common deep-learning architectures, and acceleration using custom CUDA code.
Modular C++ Monte Carlo ray tracer with subsurface scattering, GPU parallelization, and a Blender plugin.
Low-level UNIX shell in C with fork/exec/waitpid, signal handling, pipes, I/O redirection, and custom command parsing.
Banaras, India