Sentinel Gemini

Inspiration

Cybersecurity is a growing challenge in the digital world, especially with account takeovers, payment fraud, and phishing attacks becoming more sophisticated. I was inspired by the idea that AI could think like a security analyst, detect anomalies, explain risks, and recommend actions in real-time — making cybersecurity understandable and proactive for both individuals and organizations.

What it does

monitors simulated user behavior, protects accounts and payments, predicts risks, explains decisions, and interacts in real-time via a judge-friendly sandbox — all safely without using real data

How we built it

Challenges we ran into

UX Integration – initially, behavior, accounts, and payments were separate modules. Unifying them into one cohesive decision flow was difficult.

Simulated Data Realism – creating enough fake users with meaningful patterns to showcase AI intelligence was time-consuming.

AI Explanation Clarity – ensuring the AI’s reasoning was understandable to non-technical judges required several iterations.

Time Constraints – balancing feature completeness, AI intelligence, and a polished demo in a limited hackathon timeframe.

Accomplishments that we're proud of

What we learned

Working on this project taught me:

Explainable AI matters – it’s not enough to detect a threat; users need to understand why something is risky.

Simulation is powerful – creating realistic fake users and behavior helps showcase AI intelligence without touching real sensitive data.

UX + AI integration is tricky but critical – features must feel connected and meaningful, not just standalone.

Predictive reasoning and what-if simulations improve trust in AI decisions.

What's next for Sentinel Gemini

Built With

• english
• gemini