I’m a Machine Learning Research Engineer II at the Media and Data Science Research Lab, Adobe. My research interests lie in computer vision, and multimodal learning - with the goal of developing truly trustworthy ML systems that are efficient, fair, private, and accessible. I graduated with a BE in Computer Science and Engineering from Delhi College of Engineering (DTU) in 2022 and developed a strong love for research as a result of several meaningful collaborations, and internships through the course of my undergrad studies. I am currently working on multimodal pre-training for complex document understanding and fairness properties of large language models.
Note: Please view CV or Google Scholar for a recent set of projects instead of the publications tab.
Slow Science – Science (esp. ML) needs to be protected from the rush ignited by the knowledge economy. One of my hobbies is to get papers rejected and tell myself that science needs time to craft difficult questions and answers and that the goal of research is to be able to ask these questions.
Reach out to me
I’m always happy to connect with people to collaborate, indugle my sci-fi fantasies or argue about football. Feel free to reach out to me via my email :) Take a look here if you would like to find out about some of the awesome work we do at a self-funded research group!
Recent News
BIRD (Fair Knowledge Distillation using Student Feedback): accepted at Efficient Systems for Foundation Models Workshop@ICML2023.Oral acceptance of our work on prompting techniques for debiasing LLMs at AAAI Student Abstract, done by interns at MDSR Can counterfactuals help us decode more fairly? Check out our most recent preprint driven by interns at MDSR. Gave a preplacement talk at DTU, discussing my research research work and interacting with students.Densely Connected Transformer for Single Image Dehazing: accepted at Journal of Visual Communication and Image Representation.One Shot Doc Snippet Detection accepted at WACV, 2023 with collaborators at Adobe.Joined MDSR Labs, Adobe, India.Censoring using Noisy Sampling for 3D point clouds accepted at ECCV 2022.Completed undergrad!Preprint: CBNS for Private Sampling for 3D point clouds with potential for deploying ML systems in sensitive environments.Preprint: AdaSplit for Low resource distributed ML systems (Split Learning)Received pre-placement offer from Adobe!Attended NIPS '21 and presented our work on Benford's Law (paper/poster)Do Neural Network Parameters also follow Benford's Law? Our work exploring interesting properties of parameter distribution accepted at NIPS'21 ML4Physics workshop!Began Research Internship at Adobe, MDSR.Began collaborations with Ayush Chopra, Camera Culture Group (Media Lab), MIT (work on Split Learning, Private CV)Began collaboration with Yannic Kilcher (ETH Zurich) (Benford's Law and NNs).Began internship with Prof. Ganesh at Indian Institute of Technology, Bombay.Joined ML Lab at Delhi Technological University, exploring work on Image Dehazing and transformers.Completed internship at IDfy, working on a signature matching service with the data science team.Reached the final phase of IAF-Mehar Baba Prize in collaboration with Adani Aerospace (the only undergrad team in the top 3), winners of the best comm architecture.Worked on the fabrication of the VTOL system for IAF Mehar Baba Competition phase one.Joined Team UAS-DTU and began working on Indian Air Force (IAF) Mehar Baba Competition. 
