Inspiration

In India, a lot of people still don’t fully understand their rights or the government benefits available to them. The information is either too complex, scattered, or not available in their language. Because of this, many people miss opportunities or depend on others for help. We wanted to build something simple that puts this knowledge directly in people’s hands using AI.

What it does

Bharat-Seva AI is a smart assistant that helps users understand laws and find government schemes they are eligible for.

You can:

Ask questions about laws in simple language Get explanations without legal jargon Discover schemes based on your profile Learn how to apply step by step

The goal is to make important information easy to understand and actually useful.

How we built it

We built the entire system on Databricks using a Lakehouse approach.

We started by collecting datasets related to government schemes and basic legal information, then cleaned and structured them using PySpark. All of this is stored in Delta Lake tables.

For the AI part, we used two main components. One is a recommendation model that suggests schemes based on user details like income, age, and location. The second is a retrieval-based system that finds relevant information and answers user questions in a conversational way.

We used MLflow to track our models and experiments, and connected everything through a simple backend API with a chat-style interface.

Challenges we ran into

One of the biggest challenges was working with unstructured and inconsistent data. Government data is not always clean or easy to use.

Another challenge was making sure the AI responses were actually helpful and not just generic answers. We had to carefully design how information is retrieved and presented.

Also, integrating multiple components like Spark processing, ML models, and retrieval systems within a limited time was not easy.