DataFrame

• 

  Dashboard

• 

  Demo Canvas flow1

• 

  Demo Canvas flow2

Inspiration

The inspiration for Dataframe lies with Jupyter Notebook. We thought that the sequential running of Jupyter Notebooks was quite limited and not conducive to a bigger picture. We also thought of analysts and programmers who want to collaborate together and discover insights on one platform.

So thus we introduce DataFrame: the flowing canvas for Data Scientists, Hobbyists and Students

What it does it do

DataFrame is a canvas for scientists, hobbyists, students and all those who love data. Picture Jupyter Notebook and Canva had a baby. Individual Python Nodes can be manually dragged into an infinite Canvas, connected, ran and analyzed as output in a beautiful, flowing manner. Presentations can be made, notes compared, and importantly context gained in this freeflow application.

How we built it

We used a combination of frameworks, tools and languages to accomplish our goals. For the languages, we chose Python and Typescript, due to language popularity. For the frameworks, we decided on FastAPI for the backend - known for its speed in setup, and React - known for its popularity and general proficiency.

Challenges we ran into

The primary challenge we faced as a team, was Git. We foolishly decided initially to work on the main branch for the sake of expediency, which resulted in horrific merge conflicts beyond imagination. Another unrelated challenge was to merge the behaviour between Pyodide, the WASM Python port, and tldraw. We worked incrementally to slowly combine the 2 concepts and finally it all clicked together. With that feature completed, extending its functionality and adding finishing touches were much easier.

Accomplishments that we're proud of

We are proud of many accomplishments that transpired over this hackathon. Firstly, we targeted an ambitious goal, which was the creation of a fullstack, deployed application with many different moving parts. While initially we had a rough start, we managed to pull it together and complete the main functionalities of our project. Running Python on the browser and combining it with an infinite canvas had many black boxes (many of which are still unopened), but through our grit we achieved what we wanted our final product to look like.

What we learned

We collectively learned a great deal when working on this project. From technological leaps like API development and framework discovery, to character advances like risk-taking and grit, we all skilled up this hackathon. There were lessons learned in regards to version control and time management, but mainly we found ourselves gaining much needed insights to how designing software is like.

What's next for DataFrame

The possibilities are endless. Whether it be parallelism, input reactivity, collaboration and much more, this project is a blank slate waiting to be expanded. More specifically, we are planning on integrating more advanced nodes such as third-party MCP, dataset retrieving, and model training. There are alternatives no doubt, but DataFrame is uniquely positioned to hold a geniune market niche for those who wish to visualize their data in a different way.

Built With

• auth0
• fastapi
• gemini
• python
• react
• sqlalchemy
• sqlite
• sqlmodel
• tailwind-css
• tldraw
• typescript
• uvicorn
• vite
• vultr