Vedang Waradpande

I’m a Senior Machine Learning Engineer at PayPal, working on graph intelligence, search, and personalization. Prior to this, I worked as a Machine Learning Engineer II at TikTok on the integrity team, and before that in Data Science roles at Faire and Razorpay, two well-known Y-Combinator-backed startups. My core areas of interest and experience are search, recommender systems, monetization, and integrity/risk/trust and safety. I work across the full ML stack — modeling, production systems, and the data infrastructure that supports them — on projects spanning social media, e-commerce, and finance.

Alongside my industry work, I’ve been active in ML research, both full- and part-time. My core interest is computational drug discovery — molecular property prediction and docking, viewed through an ML lens — an area I find compelling for both its unsolved ML challenges and its potential real-world impact, working with Prof. Joel Freundlich and previously with Prof. Kwoh Chee Keong. I’ve also worked on research in computational genomics with Dr. Premal Shah, and in NLP and vision-language modeling with Dr. Karl Stratos.

I finished my master’s in Computer Science from Rutgers University with a Machine Learning concentration in May 2023, and my bachelor’s in Computer Science from Birla Institute of Technology and Science (BITS), Pilani (Goa campus) in May 2019.

Resume

Download my resume here

Writing

I have 2 papers under preparation, and will soon provide links to preprints here.

Articles

1. “Identification of Antituberculars with Favorable Potency and Pharmacokinetics through Structure-Based and Ligand-Based Modeling” (2025). Vedang Warapande, Fanwang Meng, Alexandra Bozan, David E Graff, Jenna C Fromer, Khadija Mughal, Faheem K Mohideen, Shivangi, Sindhuja Paruchuri, Melanie L Johnston, Pankaj Sharma, Timothy R Crea, Reshma S Rudraraju, Amir George, Camilla Folvar, Andrew M Nelson, Matthew B Neiditch, Matthew D Zimmerman, Connor W Coley, Joel S Freundlich. (bioRxiv). [PDF]
2. “Predicting Completeness of Unstructured Shipping Addresses Using Ensemble Models” (2021). Vedang A. Waradpande, Petchetti Vinay Surya Prakash, Nikhil Jhaveri, and Shashank Agarwal. In Proceedings of ACM SIGIR Workshop on eCommerce (SIGIR 2021 eCom). [PDF]

Undergraduate thesis: “Applications of Deep Graph Matrix Completion Models in Bioinformatics” (2018) [Undergraduate thesis] [PDF]

Posters

1. Graph Convolutional Neural Networks (2019). Unpublished. [PDF]

Blog

See my blog posts here.

Selected Projects

Following are some personal and academic projects I have pursued in the past couple of years related to Machine Learning, NLP, Computer Vision, and Software Engineering.

1. Visual Question Answering with Vision-Language Models
   [Code], [Paper]
   Adapts the GIT model by Microsoft for the Visual Question Answering task and evaluates it on the challenging AOKVQA dataset by the Allen Institute for AI. The dataset is designed to test the model’s ability to reason using diverse knowledge sources and commonsense.

2. Unsupervised Passage Retrieval for Fact-Checking and Entity Linking
   [Code], [Paper]
   Builds on the Unsupervised Passage Retrieval method by extending its formulation for Fact-Checking and Entity Linking tasks as well as evaluating it’s performance on the Question Answering datasets from the KILT benchmark by Meta.

3. IQ Tests on Pretrained Vision Transformers
   [Code], [Paper]
   Evaulates and improves the performance of the pretrained Vision Transformer and BEiT models on the Raven’s Progressive Matrices IQ test using the I-Raven dataset.

4. Gated Graph Neural Networks for Text Classification
   [Code], [Paper]
   Re-implementation of the TextING model using PyTorch for ease of use. Converts text to graphs and builds a Gated Graph Neural Network model for text classification and evaluates it on four standard datasets, closely replicating the original results.