Jun 2019 – Dec 2019 · 6 mos · Greater San Diego Area

• Research and development of semi-supervised and unsupervised methods for
• depth estimation in images and videos in the wild.
• Proposed a joint supervised, unsupervised and weakly supervised learning
• framework for monocular depth estimation, accepted in CVPR2020.

Mar 2019 – Jun 2019 · 3 mos · Greater San Diego Area

• CSE12: Basic Data Structure and Object-Oriented Programming

Aug 2018 – Jun 2020 · 1 yr 10 mos · Greater San Diego Area

• Developed state-of the art semi-supervised and unsupervised learning algorithms to understand depth, motion and semantic information in videos for autonomous driving car. My master's thesis was on "Learning Augmentation Policy Schedules for Unsupervised Depth Estimation"

Aug 2016 – Aug 2018 · 2 yrs

• Worked in Applied Machine Learning (AML) group. Designed and implemented
• distributed learning software pipeline in Caffe2 for large-scale training on images which reduced training time from days to hours.
• Worked on project Robocodes, which received significant media attention and
• awards. Responsible for algorithmic research on designing methods
• to extract useful information from satellite images.

Dec 2015 – Aug 2016 · 8 mos · Mumbai Area, India

• Worked as lead engineer building software systems driven by predictive intelligence
• for logistics in supply chain.
• Developed machine learning algorithms for recommending shipper for a particular
• order, predicting delivery date, last mile date, pickup date by shipper for an
• order. Skills gained in Data Analysis, Data Munging, Data Visualization, Feature
• Engineering, Feature Selection, etc.

Jun 2015 – Aug 2015 · 2 mos · Greater Pittsburgh Area

• Developed a multi-scale deep convolutional neural network for semantic labeling
• of 2D scenes for scene understanding.
• Worked on to modify this method for semantic classification of point cloud data
• forthe purpose of perception and path planning for a UAV.

Dec 2014 – Dec 2014 · 0 mo

• Created Comic Polyglot, a neural network based system that translates raw
• mangas into English, hopes to remove dependency on scanlation groups.
• Used a combination of ANN & MSER to detect text-regions with subsequent
• translation with an OCR-engine

Oct 2012 – May 2016 · 3 yrs 7 mos · Delhi Technological University, Delhi

• Developed HistExtract, a system that extracts highly correlated text from historical
• inscription images. Work featured in ”Digital Heritage” project sponsored by
• Department of Science & Technology, Govt. of India.
• Developed an autonomous navigation system using Kinect for a mobile robot.
• Project mentor to junior undergraduate students in computer vision,robotics &
• embedded system.
• LUNABOTICS MINING PROJECT:
• Project sponsored by college to make a lunabot (lunar regolith mining bot)to participate
• in NASA Lunabotics Mining Competition 2013.
• Worked in a team and developed a variety of embedded solutions like design and
• fabrication of various electronic circuits, designing the telemetry system, digging
• & dumping mechanism.