💡 Inspiration
Older generations and vulnerable individuals are constantly targeted by complex phishing and financial scam tactics. We wanted to build an intelligent defense shield that not only detects threats in real-time but also explains them to grandparents in their local languages, while ensuring an optimized and robust local memory model.
⚙️ How We Built It
The application is built entirely using Python and Streamlit for a clean intuitive interface. The AI pipeline integrates the Alibaba Cloud Qwen LLM topology for processing multilingual threat audits. To meet the Track 1 requirements, we engineered a continuous local memory layer using MongoDB Compass that captures structural signatures and intercepts repetitive patterns directly on the local machine before hitting the cloud.
🛑 Challenges We Faced
During development, we faced occasional network packet drops and connection reset issues with external endpoints. To solve this, we implemented strict defensive engineering and local failover exception handling. If the cloud engine hits a glitch, the app remains fully resilient and serves safe local mock alerts instead of crashing.
🎓 What We Learned
We mastered context lookup mechanics, learning how to link an active NoSQL database layer with cloud LLM pipelines to optimize tokens and build crash-resilient autonomous software.
Log in or sign up for Devpost to join the conversation.