Data Flow Canvas
Canvas for Data Science
2 followers
Canvas for Data Science
2 followers
DataFlow Canvas is a visual data science platform where your data never leaves your browser
š Privacy first: All processing runs locally via WebAssembly zero server uploads
š§© 300+ drag and drop blocks: transforms, ML models, statistics, and visualizations
š Full Python power: pandas, NumPy, scikit-learn no installation needed
š„ Real time collaboration with hidden IP addresses
š¦ Export pipelines as standalone Python code Open source AGPL 3.0 commercial licenses available for proprietary use
Hey Product Hunt! š
I built DataFlow Canvas because I was frustrated with how data analysis tools handle privacy. Every platform wants you to upload your data to their servers but what if you're working with sensitive customer data, financial records, or healthcare information?
The problem I wanted to solve: make powerful data science accessible to everyone WITHOUT forcing them to trust a third party with their data.
My approach evolved significantly. Initially, I thought "browser-based" meant limited capabilities. Then I discovered Pyodide Python compiled to WebAssembly. Suddenly I could run pandas, NumPy, and scikit-learn entirely in the browser. No servers needed.
The result: 257 specialized blocks covering everything from basic CSV transforms to advanced ML models like XGBoost, clustering, and time series forecasting. All running locally.
Some things I'm proud of:
ā¢ Zero data ever touches a server
ā¢ Real-time collaboration that hides your IP from teammates
ā¢ Export your pipeline as standalone Python code no vendor lock-in
ā¢ Completely free and open source (AGPL-3.0) with commercial licenses available for proprietary use.
I'd love your feedback especially on what blocks or features you'd want to see next. Data scientists, analysts, and anyone who works with spreadsheets: this is for you.
Try it at dataflowcanvas.com and start building.