Featured Security Project

Agentic Honeypot

An AI-powered honeypot that detects high-risk scam behavior, engages attackers with an autonomous persona, and exports structured threat intelligence for follow-up analysis.

Role Full Stack Builder

Category Security + AI

Status Hackathon Build

Tech Stack

Python FastAPI OpenAI API JavaScript HTML CSS

Problem

Scam conversations are usually deleted quickly, making it hard to collect repeatable evidence. Manual response handling also consumes too much analyst time for low-value interactions.

Solution

Built message risk scoring with signal detection to classify suspicious patterns.
Added an autonomous responder persona that safely delays attackers and gathers more context.
Implemented extraction for bank numbers, UPI IDs, emails, phone numbers, wallets, and URLs.
Added API export in JSON/CSV for handoff to investigation workflows.

Outcome

The system demonstrates practical value for threat-intelligence collection: lower manual effort, higher evidence quality, and a reproducible engagement flow suitable for demo and further production hardening.