Inspiration
The 1st task idea - data processing using natural language. We instantly knew it was our way to go with all possible features to implement
What it does
Well, it does what's said to be done - we developed a tool to visualize and process large amounts of data, which user can take from Open Source or upload their own datasets and get to work right ahead. Out tool using large language models to process natural human language and translate it to comprehensible instructions for computer to handle all the rest. Create views from existing data and use many of supported charts in our tool!
How we built it
We are using combination of Python FastAPI on backend and React on frontend. User can pull data from PC or from online services like World Bank Data, UN Data, OECD Data etc using different API's and merging it into one functional interface. We store data using in-memory database like SQLite for data caching. Then this cached data is processed by GPT-4o to convert understandable human prompts into SQL queries to get insights from it.
Challenges we ran into
We got in trouble while trying to visualize complex multidimensional data. We've come up with some solutions to tackle, but there's more to do in the future.
Accomplishments that we're proud of
We managed to implement key features mentioned in task 1 description
What we learned
That was our first experience using a such an amount of API's, especially using OpenAI API's to convert prompts into queries, and it was a lot easier than we expected.
What's next for Sibash-1
Who knows?
Log in or sign up for Devpost to join the conversation.