Unmasque: Shedding Light on Opaque Application Queries

Title of the Talk: Unmasque: Shedding Light on Opaque Application Queries
Speakers: Kapil Khurana Das
Host Faculty: Anupam Sanghi
Date: Feb 10, 2026
Time: 11:00 AM
Venue: LHC-8, Lecture Hall Complex

Abstract: This talk explores the emerging challenge of Hidden Query Extraction (HQE)—the task of reverse-engineering SQL queries embedded within opaque database applications. We introduce UNMASQUE, an active learning-based, non-invasive algorithm that reveals hidden warehouse queries by analyzing only the outputs of strategically mutated and generated input data. UNMASQUE is designed to be efficient and application-agnostic, making it ideal for use cases like legacy system recovery and query rewriting. We’ll touch upon its architecture, optimization strategies, and performance on real-world applications. To conclude, we’ll briefly touch on Xpose, a recent advancement that extends HQE to handle complex query structures using a hybrid of UNMASQUE and forward engineering powered by large language models.

Bio: Kapil Khurana is a Senior Software Engineer at Microsoft, where he leads initiatives in Azure SQL DB Resource Governance, focusing on overbooking, distributed resource management and multi-tier cache reclamation. He holds a Master’s degree in Computer Science from the Indian Institute of Science (IISc), Bengaluru, where he worked under the guidance of Prof. Jayant Haritsa on his thesis titled “Unmasking Hidden SQL Queries.” His research led to the development of UNMASQUE, a non-invasive SQL query extraction tool, and has been published in top-tier venues like PVLDB and ACM SIGMOD. He is also an inventor on multiple patents in database systems and machine learning integration, bridging academic rigor with real-world impact in cloud-scale data platforms.