Revathi Karri

Junior Researcher, LUT University
Research Area: Trustworthy, Explainable, and Responsible AI-Driven Software Vulnerability Detection

I hold a master’s degree in Software Engineering and Digital Transformation from LUT University and a bachelor’s in Computer Science from RGUKT Nuzvid, India. Between my studies, I worked in IT, specializing in infrastructure support and middleware administration. This experience immersed me in corporate environments and industry practices, which now enrich my research approach.

Outside of my academic and professional life, I enjoy engaging in a variety of activities that help me relax and recharge like Cooking, going for long walks, listening to music and reading bible.

Areas of Interest

AI, ML, LLM, Responsible AI, Cyber security

My Research

My research focuses on advancing AI-driven Software Vulnerability Detection (SVD) with an emphasis on creating systems that are trustworthy, explainable, and responsible. As software becomes increasingly reliant on third-party libraries and code reuse, the risk of vulnerabilities has grown. My work seeks to address these challenges by developing solutions that improve detection precision, minimize human effort, and adhere to responsible AI standards. Key areas of interest include:

• Large Language Models (LLMs) and Deep Learning (DL): Integrating cutting-edge deep learning and LLM techniques for enhanced semantic analysis in SVD, improving detection capability with minimal false positives and negatives.
• Computational Design Science (CDS): Applying CDS to develop novel algorithms and evaluate their performance rigorously through synthetic and real-world datasets, ensuring practical applicability and ethical implementation of AI in software security.
• Explainable AI (XAI) and Responsible AI Practices: Ensuring transparency, accountability, and fairness in AI-driven decisions, with a focus on reducing biases in training data for equitable vulnerability detection across diverse software ecosystems.

Current Work

Currently, I am working on Computational Literature review to identify, evaluate, and synthesize all relevant research related to AI-based SVDs, focusing particularly on studies that incorporate LLMs and DL techniques and explainability. Instead of a traditional manual review, employ computational tools and techniques to analyze the vast body of research on AI-based SVD systems. Utilize tools like text mining, natural language processing (NLP), and data visualization techniques to automate the collection and analysis of literature from large databases.

Future work

In future, I will work on the development of algorithm for improving detection capability and explainability using LLM and DL etc.,

---

Thanks for reading, if you are interested in my work, Let’s connect.