I am an M.S. in Electrical Engineering student at Stanford University and my research interests include interpretable machine learning, deep learning and NLP. For the last 2 years, I have been working at Microsoft as a Data and Applied Scientist in the Cybersecurity research team. Previously, I graduated with a Bachelors degree in Electrical Engineering and a minor in Deep Learning from the Indian Institute of Technology (IIT) Madras. During this time, I pursued research at the intersection of NLP and deep learning that led to publications in top conferences such as ACL, COLING and ALENEX.
Honors & Awards
Dr. Dilip Veeraraghavan Memorial Award, IIT Madras (2021)
Best Paper Honorable Mention, ACM CODS-COMAD (2021)
KVPY Fellowship, Government of India (2017)
NTSE Scholarship, Government of India (2015)
Education & Certifications
B.Tech, Indian Institute of Technology Madras, Electrical Engineering (2021)
Data and Applied Scientist, Microsoft (June 2021 - September 2023)
• Worked as a researcher at the intersection of data science and cybersecurity, with a focus on OAuth cloud app security.
• Developed 10 industry-first machine learning solutions spanning knowledge graphs, anomaly detection, computer vision, and NLP to model cyber attack patterns, track app behavior, and avert security threats.
• Built & deployed models that analyze terabytes of data every day meeting stringent goals on latency and efficacy.
• Filed a patent, published a paper at MLADS 2022 & received an early promotion for exceptional work.
Data and Applied Scientist Intern, Microsoft (May 2020 - July 2020)
• Developed CNN and Transformer-based deep learning models to analyze multi-spectral satellite images for estimating biomass in agricultural fields and identifying prospective areas for oil exploration.
• Designed a data structure for the open-source package, xarray to support tree-based hierarchical data storage.
Data Scientist Intern, GE Healthcare (May 2019 - July 2019)
• Used graph-based keyword clustering and topic ranking to analyze text in service records of healthcare machines.
• Set up an automated pipeline to flag common failure patterns and suggest quality improvement opportunities.
• Reduced the time taken to extract insights from service records by 11x and was appreciated by company leaders.
Input-specific Attention Subnetworks for Adversarial Detection
ASSOC COMPUTATIONAL LINGUISTICS-ACL. 2022: 31-44
View details for Web of Science ID 000828767400004
- Vocabulary-constrained Question Generation with Rare Word Masking and Dual Attention ASSOC COMPUTING MACHINERY. 2021: 431