Viraj Prabhu

Since Fall 2016, I am a research assistant in the Machine Learning and Perception lab at Virginia Tech, advised by Prof. Dhruv Batra and Prof. Devi Parikh.

My broad interests lie in computer vision and natural language processing, particularly in high-level scene understanding and multimodal learning tasks. I am interested in developing AI systems that can emulate the remarkable human ability to combine sensory inputs to develop a semantic understanding of the natural world. I'm also interested in building more interpretable AI systems that do the right things for the right reasons.

At Virginia Tech, I am presently working on applying deep learning approaches to problems at the intersection of computer vision and natural language processing. Prior to this, I worked as a front-end and Android developer at Adobe Systems India on Adobe Captivate Prime, a newly launched SAAS-based Learning Management System for enterprise.

I graduated with a Bachelor's in Computer Science from BITS Pilani in July 2015. Over the course of my undergraduate studies, I was fortunate to have the chance to work on several interesting projects in computer vision and machine learning. For a detailed list, please take a look at my CV.

I am an open-source contributor to CloudCV and served as a mentor for Google Summer of Code '16. I also occasionally dabble in writing, playing music, long-distance running and quizzing.



Research Scholar, Machine Learning and Perception Lab, Fall 2016 - Present

Advisors: Prof. Dhruv Batra and Prof. Devi Parikh

Working on deep learning approaches to Visual Question Answering.
I was also a teaching assistant for Intro to Machine Learning in Fall 2016, taught by Dr. Stefan Lee for which I developed Kaggle competitions and curated and graded homeworks and poster sessions.

Mentor, CloudCV, Google Summer of Code '16, Apr-Aug 2016

Mentored and presently maintain the Google Summer of Code project CloudCV-IDE, a web platform to build deep neural networks via a simple drag and drop interface. [GitHub]

Member of Technical Staff, Adobe Systems India, Jul '15-Aug '16

Worked on the Android and front-end webapps of Adobe Captivate Prime, a newly launched Learning Management System for enterprise.

R&D Intern, Tonbo Imaging, Jan-Jun '15

Computer vision startup developing sensor systems for battlefields and reconaissance.

Developed algorithm for automating camera calibration and implemented a boresighting algorithm for the company's video precision boresight tool.[Report]

Summer Intern, Adobe Presenter Video Express, May '14-Aug '14

Developed a graphcut-based segmentation algorithm for real-time background substitution in video that combined color, motion and shape cues and demonstrated robust segmentation across various backgrounds.[Demo][Blog][Report]

Summer Intern, May-Jul '13

Developed a web portal using the LAMP stack to automate customer data log creation for internal quality assessment purposes.

Birla Institute of Technology and Science (BITS), Pilani, Aug '11-Jul '15

Bachelor's Degree: Computer Science

Relevant Coursework: Pattern Recognition, Machine Learning, Information Retrieval, Parallel Computing

Selected Projects

Teleconferencing Using Multiple Kinects, Jan '14-Apr '14

Advisor: Dr. Jagdish Raheja, Senior Scientist, CEERI Pilani

Developed multithreaded C# application to interface multiple Kinect sensors to cover a field of vision as part of a modern teleconferencing system. Kinect Skeletal Tracking and OpenCV face detectors were used to identify and display current speaker on a central screen.[GitHub]

Topic based news aggregator, Sep '14 - Dec '14

Course project for Information Retrieval. Integrated a Python web crawler with a hierarchical agglomerative clustering algorithm to fetch, identify and chronologically present news articles pertaining to the same event. [GitHub]

Sign Language to Speech Converter, Dec '13-Feb '14

Trained a Hidden Markov Model to recognize American Sign Language gestures on image features extracted using the Kinect’s Skeletal Tracking libaries and OpenCV. Presented at APOGEE 2014, BITS Pilani's annual all-India technical symposium.

Branch and Bound, Feb '14-Apr '14

Course project for Parallel Computing. Implemented algorithms for Travelling Salesman and Knapsack problems using a Branch-and-Bound framework and parallelized them using OpenMP and MPI.[GitHub]


I enjoy developing chrome extensions for applications I find useful. I developed Read Timer to provide an estimate of the time it will take to read an article, and plot reading speed over time. Another app was quickmark, a tool to bookmark efficiently using typeahead and keyboard shortcuts. At Angelhack '16, we built Omni, an indoor navigation system.

Awards, Positions and Extra-curriculars

  • Second Place (Adaptive Technologies Category, 35+ entries) in Project Presentation, APOGEE 2013 for MARS, a genre based music equalizer that used an SVM classifier trained on MFCC features.

  • Second Place (Design Appliances Category, 40+ entries) in Project Presentation for Try-on, an application that used the Microsoft Kinect to accurate measure a user’s clothing size using the Kinect’s Skeletal Tracking libraries.

  • Winner, Google Hackathon, APOGEE 2014, for developing Snapify, an image-sharing Android app.

  • GSoC Mentor Summit 2016: I represented CloudCV at the mentor summit in Sunnyvale, CA

  • Editor, International Press during BITSMUN 2013, an all-India Model United Nations conference where I led a team of 25 student reporters, photographers and designers

  • Head Boy, Lilavatibai Podar High School from ’09-’10. I represented my school at intercollegiate competitions and events.

  • Writer for the English Press Club and member of the Department of Sound at BITS Pilani

  • Other Interests: Football (member of Adobe Football team), Long distance running, Piano performance

Contact me

Primary email: username -at- (where username = virajp, domain = vt)

You can also reach me on any of the following: