dhruvsrikanth / model-pipelines Goto Github PK

I am a researcher at Weco AI, a startup in London building for devs. Previously I spent time at the Auton Lab at the Robotics Institute at CMU, where I worked on deep learning and algorithms for computer vision, robotics and time-series. Previously, I've worked on deep learning and algorithms for fairness, robustness and generalization in computer vision and 2D-3D single and multi-view reconstruction. I've spent some time doing applied machine learning and deep learning research at a few startups, GE Healthcare and GE Aerospace.

Apart from AI research, I enjoy high-performance computing and all the kernel/thread/cluster programming (debugging lol) that comes with it. I've made a few games, simulations, and systems.

I like to apply my research to healthcare, climate, industrial, and scientific problems. I am interested in AI and research at companies that take a practical engineering approach to innovation and impact.

Outside of research I enjoy being outdoors and playing sports like tennis, golf and football.

           

Deep Learning
AutonFeat: A high-performance Python package for automatic and distributed time series feature extraction and selection
tsfresh: A time series feature extraction package
Denoising Diffusion Probabalistic Models (DDPMs)
2D-3D Single View Reconstruction
pynn (Deep Learning Framework with pure NumPy Computation)
VanillaGAN (Vanilla GAN experimentation library)
ConditionalGAN (Conditional GAN experimentation library)
DCGAN (DC GAN experimentation library)
PSNRGAN (Contrastive PSNR GAN experimentation library)
Monocular Depth Estimation
Neural Artistic Recreation (A CLI for Neural Style Transfer)

High-Performance Computing
CUDANN (Distributed Deep Learning Framework accelerated on NVIDIA GPUs)
FastConv (Multithreaded GPU CUDA kernels for Convolution)
Ray Tracing (using Monte-Carlo Algorithm on GPU Accelerators)
Advection Diffusion Simulation (Multithreaded, Shared Memory Parallelism, Distributed Memory Parallelism)
N Body Particle Simulation (Multithreaded, Shared Memory Parallelism, Distributed Memory Parallelism)
Distributed Poisson Solver (using Conjugate Gradient Optimization)
Image Editor via Convolutions (using Work Stealing and Work Balancing parallel schemes)
Sparse Linear Solver (using the Map Reduce parallelism scheme with Conjugate Gradient Optimization)
Image Editor via Convolutions (using Pipeline and BSP parallel schemes)
Twitter Application (using Producer-Consumer parallelism model)

Machine Learning
Boosting and Boostability (An understanding through Computational Learning Theory)
Financial Forecasting of TSLA (using sentiment analysis, feature engineering, and RNNs)

Computer Systems
GoLLUM - A GoLang Compiler
Unix Shell

Games
Climbing Mount Everest (A text-based adventure game)

Other
Useful Algorithms
Superconductor Regression (Kaggle Challenge)
Canvas Web Application
Blood Donations Database