Projects

Whispers of Wealth: Red-Teaming Google’s Agent Payments Protocol

Conducted AI red-teaming research on Google’s Agent Payments Protocol (AP2) by building a functional agent-based prototype and performing adversarial prompt-injection testing to evaluate security risks in conversational payment systems.

Identified vulnerabilities including:

• Unauthorized tool invocation
• Data exfiltration
• Manipulation of payment authorization flows

Designed Branded Whisper and Vault Whisper attacks and proposed mitigation strategies for safer agentic commerce systems.

Research Areas:
Agentic AI, AI Security, Prompt Injection, Financial Agents, Red-Teaming, LLM Security

Paper Link

CREaiT: Human-Centered AI for Creative Communities

Designed and evaluated a human-centered AI system enabling:

• Artist opt-in data sharing
• Transparent AI credibility signaling
• Fair compensation mechanisms

Conducted semi-structured interviews with artists, researchers, and industry practitioners and translated findings into UX concepts, wireframes, and prototypes.

Research Areas:
Human–Computer Interaction, Human-Centered AI, Ethical AI, UX Research

ArchLibrary: Library Management System

Developed a full-stack library management system for UGA using:

• Spring Boot
• MySQL
• JWT Authentication
• Next.js

Implemented:

• Book borrowing system
• Fine management
• Transaction tracking
• Role-based access control
• RESTful APIs

Technical Areas:
Database Systems, Full-Stack Development, REST APIs, Authentication

Information Retrieval Search Engine based on Tf-IDF

Implemented three functions: Calculate token IDF, document-wise normalized TF-IDF weight, and cosine similarity score for a query-document matrix using the US Inaugural Addresses dataset.

Research Areas:
Text Mining, Tf-IDF, Natural Language Processing, Information Retrieval

LLM-based Q&A Chatbot for E-Learning

Built an end-to-end LLM-based Q&A system for an e-learning company using:

• Google PaLM
• LangChain
• Streamlit
• Text embeddings

The system helps students retrieve course-related answers efficiently and improves learning support workflows.

Technical Areas:
LLMs, NLP, LangChain, Streamlit, AI Applications

Facial Expression Recognition with Deep Learning

Implemented facial expression recognition using:

• CNN
• Transfer Learning
• SVM
• KNN
• Random Forest

Improved performance through:

• Image preprocessing
• Feature extraction
• Hyperparameter tuning

Technical Areas:
Computer Vision, Deep Learning, Machine Learning

Integrated Directory Synchronization and Management System

Designed and implemented a distributed directory synchronization system with one client and two servers to support real-time merging, sorting, and data synchronization, along with document management features such as file locking, task queuing, and user notifications.

Research Areas:
Distributed Systems, Client–Server Architecture, Concurrency Control, Real-Time Synchronization