Hey there! 👋🏼
Welcome. I'm a maker and a ML Engineer.
I like building fun and exciting products. I've recently taken on a challenge of building 21 products in 2021 (or approx 2 products each month ). Work at Dunder Mifflin is my first completed project in this challenge.
I also run FalconML on the side where I offer end-to-end ML advisory and services.
Check out some stuff I've built
Let's connect! Feel free to reach out. I won't bluetick :)
stuff I've built
Built an ML model to segment Mumbai Slums. Fun fact, 70% of Mumbai are slums. The stark dichotomy of massive skyscrapers and slums nearby motivated me to make this. I wanted to find out how much slums have grown over time.
More drops incoming.
Human Pose estimation is an important problem that has enjoyed the attention of the Computer Vision community for the past few decades. In this post, I write about the basics of Human Pose Estimation (2D) and review the literature on this topic
Ranks in top 3 in Google SERP for 'Pose Estimation'. Recommned reading as part of UMD MS CS coursework.
In this post, I write about the basics of 3D Human Pose Estimation and review the literature on this topic.
3D pose estimation is the task of producing a 3D pose that matches the spatial position of the depicted person.
Fellow |Singapore | July 2019 - Dec 2020
Selected to be part of the 6th cohort of EF Singapore.
EF is backed by Reid Hoffman (founder of LinkedIn), founders of DeepMind and Paypal, Greylock Partners, Mosaic Ventures and some of the world's other top investors. Since 2011 EF has worked with 2000+ individuals to build over 300+ startups that are valued in excess of $2B
Machine Learning Engineer
Bangalore | Jan 2019 - July 2019
Mumbai | August 2018 - Nov 2018
Here I reaslised I don't like academic research.
I don't really do research and papers anymore. You can check out my previous work if you're interested.
View my google scholar page here. My H-index is 2 and I have 12 citations.
NeurIPS ML4D Workshop, 2018
We introduce an approach to segment and map individual slums from satellite imagery, leveraging regional convolutional neural networks for instance segmentation using transfer learning. In addition, we also introduce a method to perform change detection and monitor slum change over time. We show that our approach effectively learns slum shape and appearance, and demonstrates strong quantitative results, resulting in a maximum AP of 80.0.
Relation Networks for Optic Disc and Fovea
Localization in Retinal Images
NeurIPS ML4 Health Workshop, 2018
We propose a novel approach to localizing the centers of the Optic disc and Fovea by simultaneously processing them and modelling their relative geometry and appearance. We show that our approach improves localization and recognition by incorporating object-object relations efficiently, and achieves highly competitive results.