Student projects I guide. Each B.Tech capstone runs over two semesters (CP1 and CP2) and must end in a complete, independently tested engineering system. Both teams below have already had a paper from their work accepted at a conference.
B.Tech capstone projects
C19 (Cyber-19): Automated Web-Application Vulnerability Scanner
Domain: Cybersecurity and Blockchain Systems · Reference: OWASP ZAP and the OWASP Top 10
Web applications have a huge attack surface, and OWASP Top 10 flaws such as injection, broken access control and cross-site scripting are everywhere. Automated scanning helps developers find them before attackers do. The team is building an educational scanner, used only on authorised targets, that detects OWASP-class flaws and explains how to fix them.
- CP1 (reference): checks for XSS, injection and access-control flaws.
- CP2 (contribution): severity scoring and remediation guidance for each finding.
- Final system: a reporting interface, benchmarked against a baseline scanner.
- Acceptance: detection accuracy and false-positive rate, measured independently.
Team: Singirikonda Saimahima, Podila Lakshmi Prasanna, Mandru Sirisha.
Paper accepted: “Protocol-Invariant Cross-Chain Attack Detection: Domain-Generalized Monitoring for Previously Unseen Bridges”, International Conference on Intelligent Networks and Data-driven Intelligent Applications (ICon INDIA 2026). Registered and presented papers will be considered for IEEE Xplore.
C20 (Cyber-20): Federated Collaborative Threat Detection
Batch: B66 · Domain: Cybersecurity and Blockchain Systems · Reference: Flower federated learning framework
Organisations detect threats better together, but privacy and competition stop them from sharing raw security data. Federated learning lets them train a shared threat detector without pooling their data. The team is building a federated detector that shows a measurable benefit over each organisation's local model while keeping model updates private.
- CP1 (reference): federated training of a shared threat detector.
- CP2 (contribution): handling non-IID data, where each organisation sees a different mix of attacks.
- Final system: private model updates and a measured collaboration benefit.
- Acceptance: an independent check that the federated model beats local models under non-IID data.
Team: Pavuluri Jeshwanth Sai, Yash Verma, Gujjula Jyothi Sai Venkata Vyshnavi, Kancharla Jo Srinivas.
Paper accepted for oral presentation: “Sybil-Resilient and Publicly Verifiable Federated Intrusion Detection for Heterogeneous IoT Networks”, 10th Conference on Information and Communication Technology (CICT 2026), with the IEEE Hyderabad Section, IIITDM Kurnool, 18–20 December 2026. Proceedings to be submitted to IEEE Xplore.