Exploring graphics in Rust with Vulkan and Vulkano

Writing a compiler for PyTorch that optmizes neural nets for NVIDIA GPUs

Throwback to a time before 4 websites ruled the Internet, and the case for bringing back source curation

Showing how you can use DRIVE AGX and its suite of accelerators to achive real-time perception for self driving cars

Reconfigurable Self Organizing Office Lighting

Making it super easy to get started with Deep Learning using Jupyter Notebooks and Docker

Making NVIDIA's Neural Network Optimizer accessible to the everyday data scientist

A vision for HackIllinois as both an organization and event as it continues to grow while the hacking community enters a time of uncertainty. We present a pivoted event to mitigate some of the current major issues with hackathons and a roadmap that will allow HackIllinois as an organization to be more effective in putting on extremely successful events and engaging both the local and greater hacking community. If properly executed, this will propel HackIllinois to becoming a thought leader in the hackathon domain as well as start to shift the hacking culture to a more sustainable set of values.

Our entry into the Anheuser Busch hackathon

We introduce a novel framework for creating unified application/programmer interface (API) layers for the microservice architecture. We show that because of this new framework called Arbor, we are able to allow a whole new class of developers to create infrastructure in the microservice paradigm and in particular allow for high churn development groups (college groups for instance) to continually update their software infrastructure, without significant additional documentation and refactoring.

A DSL made by Nest Labs to easily develop end-to-end IoT services

A method for compressing images, prioritizing semantically important data

We propose an approach for 3D reconstruction and segmentation of a single object placed on a flat surface from an input video. Our approach is to perform dense depth map estimation for multiple views using a proposed objective function that preserves detail. The resulting depth maps are then fused using a proposed implicit surface function that is robust to estimation error, producing a smooth surface reconstruction of the entire scene. Finally, the object is segmented from the remaining scene using a proposed 2D-3D segmentation that incorporates image and depth cues with priors and regularization over the 3D volume and 2D segmentations. We evaluate 3D reconstructions qualitatively on our Object-Videos dataset, comparing to fusion, multiview stereo, and segmentation baselines. We also quantitatively evaluate the dense depth estimation using the RGBD Scenes V2 dataset [Henry et al. 2013] and the segmentation using keyframe annotations of the Object-Videos dataset.

Simple utility to enable quick collaboration

Creating UIUC's freshman CS Projects Course

A TEDx Talk I gave recounting my experiences needed to apply ideas from one field into another

My first published work on developing user interfaces that span multiple independent nodes in communication in a smart wall

The classic Google Billboard problem, find the first 10 digit prime number in the digits of e