Vizier

Inspiration

Data is all around us, but getting valuable insights is time consuming.

What it does

Take a description of your desired dataset, parses it into a json schema and generates sensible Plotly dashboards from that data. Allows you to populate the graphs with mock data or real data samples.

How we built it

Frontend was built with Vite and Preact in Typescript. Backend is built in Python with FastAPI. Created with the help of AI coding agents through OpenCode.

Challenges we ran into

The LLM can sometimes output incomplete data. Tried to minimize it with prompt refinement.

Accomplishments that we're proud of

Full end-to-end flow with real data

What we learned

AI coding agents allows you to build applications from scratch at an immense speed.

What's next for Vizier

More polish, options to extract the generated graphs and to make them more interactive by adding filters etc.

Built With

• mantine.js
• opencode
• preact.js
• python
• tailwind-css
• typescript
• vite