Disastra

Website Interface

Disastra (derived from the Sanskrit word for "Disaster") is a web-based application designed to help predict natural calamities like cyclones and earthquakes, leveraging the power of Machine Learning (ML) and Artificial Intelligence (AI). It was built using Python for backend intelligence, React.js for the frontend interface, and integrated ML models to process and analyze environmental data.

The inspiration came from the increasing frequency and severity of climate-related disasters. Often, the lack of timely prediction or awareness leads to loss of life and property. With Disastra, we wanted to build something meaningful, A tool that provides early insights and real-time risk alerts to help communities stay prepared.

What I Learned from the Project:

Machine Learning Implementation: Learned how to preprocess meteorological and seismic datasets, train models to classify disaster risk levels, and make predictions based on geolocation. React.js Frontend Development: Built a responsive UI that adjusts seamlessly across desktop and mobile devices. Geolocation & API Integration: Integrated location-based services and real-time data fetching using APIs for accurate predictions. User Authentication Flow: Implemented secure user registration and login with location access permissions.

How its build:

Frontend: Built using React.js, designed to be clean and responsive. Backend: Developed with Python and Django REST Framework, handling data collection, ML predictions, and user authentication. Machine Learning: Used historical cyclone and earthquake data to train classification models (like Random Forest and Logistic Regression). APIs: Connected to open datasets and APIs for real-time weather/seismic data. Charts & Visualization: Displayed prediction trends and alerts using libraries like Chart.js.

Challenged Faced:

Data Collection & Cleaning: Obtaining reliable and high-quality disaster data was difficult. Many sources had inconsistencies or missing values that needed extensive cleaning. Model Accuracy: Fine-tuning the ML model to avoid false alarms while maintaining sensitivity to real risks was a major hurdle. Geolocation Handling: Ensuring the system smoothly requested and used user location data securely across browsers and devices. Deployment Issues: Integrating backend APIs and ensuring cross-origin requests worked properly when converting the project into a mobile-friendly APK.

Disastra helps users:

Stay informed with real-time disaster risk alerts.
View dynamic graphs and risk charts based on their location.
Benefit from a seamless experience that combines AI, open data, and intuitive design.

Built With

Submitted to

World’s Largest Hackathon presented by Bolt

Created by

Disastra inspired by the Sanskrit word for “disaster” is a visionary web-based platform designed to anticipate natural calamities like cyclones and earthquakes using the transformative power of Machine Learning (ML) and Artificial Intelligence (AI). With a Python-powered backend, a seamless React.js frontend, and deeply integrated predictive models, Disastra analyzes real-time environmental data to forecast potential threats before they strike.

The driving force behind Disastra was not just technology it was humanity. As climate-related disasters grow in both frequency and intensity, we’ve seen how entire communities can be blindsided, suffering loss, trauma, and devastation. We built Disastra to be more than just a tool it’s a digital guardian. A platform born from empathy, purpose, and a desire to help people stay one step ahead of nature’s fury. By offering early insights and real-time risk alerts, Disastra empowers individuals, families, and governments to act swiftly, protect lives, and build resilience in the face of uncertainty.

What I do

As part of the Disastra project, I implemented scheduled notifications using Python’s schedule module in Django. This functionality enables the system to deliver timely alerts to users through a connected API, ensuring they receive real-time disaster risk updates. I was actively involved in both the development and testing phases of this feature to ensure reliability and responsiveness.

Beyond the backend, I also contributed to the project’s presentation by editing a demonstration video highlighting how Disastra works, its impact, and the technical workflow behind it. This helped effectively communicate our vision and the platform’s capabilities to a wider audience.

The EMon
This project was solely developed by me from ideation and design to frontend, backend, and machine learning integration. I handled every aspect of development, including writing the code, training models, designing the UI, setting up the backend infrastructure, and deploying the final product.

While I built the project independently, I’m incredibly grateful for the support, motivation, and encouragement from my friends and family, which kept me focused and inspired throughout the hackathon journey.

Brijesh Yadav