A collection of things I've built, spanning quick experiments, hackathon sprints, and more deliberate systems. Each project reflects a different idea explored, problem tackled, or boundary pushed, brought to life through constant iteration and hands-on building.
Personal
01
A comprehensive automated pentesting framework that uses a multi-phase AI workflow to handle discovery and reconnaissance, intelligent test planning, dynamic payload generation, and automated execution targeting all web vulnerabilities. The system features adaptive exploitation powered by LLM agents with goal-oriented testing across complete attack surfaces.
02
An AI-powered static analysis tool that automatically detects security vulnerabilities in source code using advanced machine learning models, pattern recognition, and deep semantic analysis. It features intelligent code parsing, vulnerability classification, risk scoring, and automated reporting with remediation recommendations.
03
A fine-tuned LLM-based transcript analyzer designed for advanced scam call detection with real-time emotion detection, voice pattern analysis, and background noise classification. Integrated into a mobile application that provides instant alerts and automated response using ML-driven pattern recognition and threat classification.
04
A SOC platform that utilizes ensemble methods like Random Forest and XGBoost alongside deep neural networks including LSTM and CNN for real-time threat detection. It features behavioral anomaly analysis using isolation forests, adaptive network scanning with TCP and UDP fingerprinting, automated incident response workflows, and MITRE ATT&CK framework integration.
05
A hackathon-built real-time city intelligence app for Bangalore that analyses the mood and vibes of different areas including weather conditions, traffic density, impactful events, crowd sentiment, and local activity. It aggregates multiple APIs and AI-driven analysis into an interactive map with live zone insights.
06
A full-scale enterprise web application for comprehensive inventory and billing operations. It features multi-user role management, real-time tracking, automated workflows, financial analytics, reporting dashboards, supplier management, and integrated payment processing with audit trails.
07
A two-agent AI assistant system where one agent handles real-time conversation while the other manages intelligent caching and conversation history storage. Built with graph-based RAG and a hybrid database architecture to give the LLM true persistent memory and human-like contextual awareness including time, weather, and environmental signals.