Dr. Komondoor V. Raghavan
Associate Professor, Department of Computer Science and Automation
Indian Institute of Science, Bangalore

TITLE: Model-checking database-accessing applications

We will present a novel approach for analysis and verification of database-accessing applications that use the ORM (Object Relational Mapping) paradigm. Our approach infers a general-purpose relational algebra summary of each user-invocable method in the application. This summary can then be fed into any off-the-shelf relational algebra solver to check for properties or specifications given by a developer. We have implemented our approach as a prototype tool that works for Spring-based MVC applications, and have evaluated it using numerous real programs. This talk is based on a paper that appeared in the proceedings of International Conference on Software Engineering (ICSE) 2022.

Dr. Atul Kumar
Senior Research Scientist, IBM India Research Lab
Bangalore

TITLE: Reimagining Software Engineering: Augmenting Classical Techniques with Generative AI

The emergence of Generative AI (GenAI) and large language models (LLMs) has begun to reshape the landscape of software engineering research and practice. From code generation to test synthesis and documentation, LLMs promise unprecedented automation and productivity gains. Yet, alongside these opportunities lie fundamental challenges, primarily the reliability and trustworthiness of AI-generated artifacts.

In this talk, I will explore how we can bridge the gap between the promise and the reality of GenAI in software engineering. I will highlight approaches that combine classical software engineering techniques — such as static and dynamic analysis, data transformation, symbolic execution — with GenAI to achieve outcomes that are both innovative and reliable. Drawing on examples from recent work on legacy Cobol/Mainframe applications, I will discuss how this hybrid paradigm can address the limitations of current GenAI models, reduce risks, and open new avenues for research. The talk will also reflect on emerging challenges and opportunities for the software engineering community as we enter this new era of AI-augmented software development.

Dr. Jyothi Vedurada
Assistant Professor, Computer Science & Engineering
Indian Institute of Technology Hyderabad

TITLE: Lightweight Models and Semantic Representations for Secure Software Development

ABSTRACT: Integrating intelligent code analysis and repair into everyday development environments has the potential to greatly improve software security, reliability, and developer productivity. However, large language models demand substantial compute resources that are often impractical for deployment on low-resource devices. In this talk, I will present our work on designing lightweight Code Language Models (CLMs) that can accurately perform tasks such as vulnerability detection, bug classification, and code summarization. By combining targeted input representations, pattern-aware encodings, and efficient classifiers, our models capture deep semantic information and overcome token length constraints, making advanced code intelligence more practical and sustainable. I will highlight Patracer, a system that integrates semantic patterns with program chunking to achieve state-of-the-art vulnerability detection, as well as AFGNN, a graph-based framework for identifying API misuses in Java code. I will also briefly share our efforts in automated security bug repair and documentation generation.