Pinkesh Badjatiya

pinkeshbadjatiya [at] gmail [dot] com | +91-720-774-6433

Resume | Google Scholar | dblp | LinkedIn | Github

I am currently working with the Media and Data Science Research (MDSR) Lab - a BU embedded research group in Digital Marketing at Adobe as a Machine Learning Researcher and Development Engineer 2, where I mainly focus on Computer Vision and Reinforcement Learning.

I am interested in artificial intelligence, natural language processing, machine learning, and computer vision. Much of my research is in Deep Reinforcement Learning (deep-RL), Natural Language Processing (NLP), and training Deep Neural Networks to solve complex social problems. I am also interested in bringing these recent developments in AI to production systems.

I finished my MS by Research in Machine Learning with a focus in the domain of Information Retrieval and Extraction (IRE) from IIIT Hyderabad under the guidance of Prof. Vasudeva Varma (former Dean R&D, IIIT Hyderabad) and Prof. Manish Gupta (Microsoft, India). My thesis was encompassing the realm of Hate Speech Detection and Bias Mitigation in neural networks. Insights from my research work subsequently resulted in the formation of Project ANGEL which later garnered the Google AI/ML Research Award 2020. Other projects I worked on during my studies include Trust Inference in Social Network, Document Segmentation, Text Summarization and Project Sexism.

I also completed my B.Tech with Honours in Computer Science and Engineering from IIIT Hyderabad.

I spend most of my time reading books. I also spend my spare time pursuing my childhood passion; following Astronomy, General Relativity, and Quantum Mechanics. Occasionally, I play piano (Youtube), listen to music, eat, drink, and sleep.

profile photo


Work Experience

Machine Learning Engineer 2
June 2019 - Current
Adobe

I primarily work with Data Science, NLP, and Reinforcement Learning research. I also work towards building ML algorithms for production.

Machine Learning Research Intern
Jan 2019 - April 2019 (4 months)
Adobe

Working towards solving social dilemmas for Multi-agent setting using Deep Reinforcement Learning using Model-based Value functions. Extended a full-time offer.

Software Engineer Intern
May 2018 - July 2018 (3 months)
Goldman Sachs

Designed an algorithmic solution using constraint optimization in java and scala. Used TDD with OOP concepts and Design patterns. Performed error analysis and deployed to production. Extended a full-time offer.

Data Analytics/Site Reliability Engineer Intern
Sep 2017 - Mar 2018 (7 months)
T-Hub Homepage

Developed analytics from the reports produced by ~280 co-working startups incubated at T-Hub. Analyzed the user traffic flows to improve the site experience and generate weekly statistics for internal-analysis using Google Analytics and Python

Google Code-In Mentor
Nov 2016 - Jan 2017 (3 months)
Google Code-In / MetaBrainz Homepage

Mentored students working on the ListenBrainz project. Tasks involved creating tasks of varying difficulty levels and evaluating their submissions.

System Administrator and Organizer
Jul 2016 - Jun 2017 (11 months)
Felicity Website

Felicity Threads is the annual technical fest of IIIT Hyderabad. I was part of the team responsible for maintaing the contest portals and the website for the fest. I was also a problem setter for the event Break-In CTF and

Google Summer Of Code (GSoC) Intern
April 2016 - August 2016 (4 months)
Google Summer of Code / Blog@MetaBrainz

Created a proxy submission Flask API compatible with Last.fm scrobblers. Added scrobbling support for desktop clients with support for tracking currently playing song in Redis.



Patents

Entropy Based Synthetic Data Generation For Augmenting Classification System Training Data
Pinkesh Badjatiya, Nikaash Puri, Ayush Chopra, Anubha Kabra
US Patent 16/659,147 (Published)
Text Conditioned Image Search Based On Dual-disentangled Feature Composition
Pinkesh Badjatiya, Surgan Jandial, Pranit Chawla, Mausoom Sarkar, Ayush Chopra
US Patent 17/160,862 (Filed)
Text-conditioned Image Search Based On Transformation, Aggregation, And Composition Of Visio-linguistic Features
Pinkesh Badjatiya, Surgan Jandial, Pranit Chawla, Mausoom Sarkar, Ayush Chopra
US Patent 17/160,893 (Filed)


Research

I'm interested in computer vision, machine learning and its applications, and natural language processing (NLP). Representative papers are highlighted.

Combating Online Hate Speech Roles of Content, Networks, Psychology, User Behavior and Others
Sarah Masud, Pinkesh Badjatiya, Dr. Amitava Das, Dr. Manish Gupta, Dr. Vasudeva Varma, Dr. Tanmoy Chakraborty
Tutorial Accepted at ECML/PKDD 2021
Tutorial-Website
Inducing Cooperative behaviour in Sequential-Social dilemmas through Multi-Agent Reinforcement Learning using Status-Quo Loss
Pinkesh Badjatiya, Mausoom Sarkar, Abhishek Sinha, Siddharth Singh, Nikaash Puri, Jayakumar Subramanian, Balaji Krishnamurthy
Under Review
Accepted as Extended Abstract in AAMAS 2020
paper
Unsupervised Hierarchical Concept Learning
Sumegh Roychowdhury, Sumedh A Sontakke, Nikaash Puri, Mausoom Sarkar, Milan Aggarwal, Pinkesh Badjatiya, Balaji Krishnamurthy, Laurent Itti
Under Review
paper
TRACE: Transform Aggregate and Compose Visiolinguistic Representations for Image Search with Text Feedback
Surgan Jandial*, Ayush Chopra*, Pinkesh Badjatiya*, Pranit Chawla, Mausoom Sarkar, Balaji Krishnamurthy
* denotes equal contribution
Under Review
paper
Leveraging Style and Content features for Text Conditioned Image Retrieval
Pranit Chawla, Surgan Jandial, Pinkesh Badjatiya, Ayush Chopra, Mausoom Sarkar, Balaji Krishnamurthy
Published at CVFAD - CVPR 2021
paper
MixBoost: Synthetic Oversampling with Boosted Mixup for Handling Extreme Imbalance
Anubha Kabra, Ayush Chopra, Nikaash Puri, Pinkesh Badjatiya, Sukriti Verma, Piyush Gupta, Balaji Krishnamurthy
Published at IEEE International Conference on Data Mining (ICDM), 2020
paper
Multi-label Categorization of Accounts of Sexism using a Neural Framework
Pulkit Parikh, Harika Abburi, Pinkesh Badjatiya, Radhika Krishnan, Niyati Chhaya, Manish Gupta, Vasudeva Varma
Published at Empirical Methods in Natural Language Processing (EMNLP), 2019
paper
Stereotypical bias removal for hate speech detection task using knowledge-based generalizations
Pinkesh Badjatiya, Manish Gupta, Vasudeva Varma
Published at The Web Conference (TheWebConf), 2019
paper
Attention-based neural text segmentation
Pinkesh Badjatiya, Litton J Kurisinkel, Manish Gupta, Vasudeva Varma
Published at ECIR, 2018
paper / code
Deep learning for hate speech detection in tweets
Pinkesh Badjatiya*, Shashank Gupta*, Manish Gupta, Vasudeva Varma
* denotes equal contribution
Published at WWW, 2017 (Best Poster Award)
paper / Media Coverage


Service

PC Chair
  • CODS-COMAD 2022
  • WOAH-EMNLP 2021
  • AAAI 2021
  • ICON 2019
Reviewer
  • ICCV 2021
  • CVPR 2021
  • AAAI 2021
  • CODS-COMAD 2021
  • WOAH-EMNLP 2020
  • ICWSM 2019
  • ICON 2019
  • ALW 2019
  • EMNLP 2018
IIIT Hyderabad
Graduate Teaching Assistant


Website template borrowed from here.
If you wish to use my version of the template, please provide credits.