victim-support

Scam awareness app, contains 6 distinct categories with each consisting of 9 yes/no questions. The results can vary between low/medium/high risk depending on what the user selects.

Evidence-first AI for digital harassment — structure evidence, add legal context and prepare a human-reviewable case package.

Technical case study of a 600+ day sophisticated Android RAT campaign involving DSU/GSI multi-environment infrastructure, baseband diagnostic exploitation, and cross-border C2

First-response agent for fraud victims — plain-language story in, deterministic evidence extraction, drafted IC3 complaint + exchange freeze letter + action plan out. Strands Agents SDK; the model never touches a number. AWS Agents for Humans hackathon.