SafeShift AI :-Next Gen migration planning AI
SafeShift AI: A Climate Change Migration and Resilience Planner
SafeShift AI is an end-to-end decision-support tool that transforms historical disaster data into actionable, real-time climate migration and resilience policy. Built over a two-month hackathon period, this solution uniquely integrates a predictive machine learning model with the advanced reasoning of the Gemini API to address one of the century’s most pressing humanitarian challenges.
The Problem Solved:
Climate-driven displacement is growing, but governments often lack the proactive tools to plan for incoming populations. SafeShift AI bridges this gap by delivering forecasts alongside ready-to-implement strategic plans.
Key Technical Components:
Time-Series Forecasting Engine:
We cleaned and engineered features (lag variables) from decades of historical Indian disaster data.
The model (built using Scikit-learn) generates a critical forecast: the predicted number of people affected by climate events in the upcoming year (e.g., 3.18 million in 2025). This number drives all subsequent planning.
The SafeShift AI Dashboard (Streamlit):
A clean, professional application built entirely in Python and Streamlit (with custom dark-mode theming).
Features key metrics, historical context, and a geospatial map visualization that highlights high-risk zones needing immediate intervention.
Strategic Planning with Gemini API (The Innovation):
This is the core value proposition. The dashboard sends the raw numerical prediction to the Google Gemini API.
Gemini is assigned the persona of a Chief Resilience Officer and generates a tailored, three-part strategy: Short-Term, Mid-Term, and Long-Term planning advice.
Crucially, this advice is generated with Google Search Grounding enabled, ensuring the policy recommendations are validated against real-time, current government initiatives and news—making the output immediately useful and reliable.
SafeShift AI is more than just a forecast; it's a blueprint for proactive climate adaptation, turning data science into lifesaving policy.
Project credits
Akkanavaru Shrawani
Saumya Belgi
Divisha Panjwani
Tanisha Chopade
Bhakti Deshmukh