I am a part of the team developing a 3D computer vision system for agricultural automation.

I have been a part of the team developing automated acquaculture solutions.

I am a part of the team working on automated construction machinery in Moonshot CAFE project.

Every AI engineer uses GPUs and CUDA, but how well do we know our GPUs? In this two-part series, .I introduce the CUDA programming model and analyze AI acceleration features in 5 generations of NVIDIA GPUs.

I am writing a series of blog posts on creating industrial grade embedded AI solutions on Jetson AGX Xavier with TensorRT. The first part is out. Read the blog post on Learn OpenCV.

Learn how to apply your computer vision skills to create industrial products. This is the second part of the industrial embedded AI series. Read the blog post on Learn OpenCV.

Home security using embedded devices ranging from the Raspberry Pi Zero all the way upto Jetson AGX Xavier. This system has been in use at my home for ~5 years. Listed as a Jetson community project by NVIDIA.

I implemented real-time style transfer from scratch in PyTorch and showed the results live in a zoom meeting. Read the blog post about this on Learn OpenCV.

I extended NVIDIA Object Detection Toolkit (ODTK) to perform object detection in real-time on a video footage. My pull request was accepted and merged into the ODTK. Results verified on drone.

I built a *safe* and powerful laser communication system capable of sending messages into space.

I implemented FlyHash algorithm proposed by Dasgupta et. al. and explained their paper.

I worked with Prof. Saket Navlakha from Salk Institute of Biological Studies to improve the performance of fruitfy inspired ANN algorithm. Our algorithm increased search mAP by 50%-90%.

I implemented Swish and wrote a blog post about it within a day after the paper came out.