// the find
h2oai/wave
Realtime Web Apps and Dashboards for Python and R
H2O Wave lets you build real-time browser dashboards entirely in Python or R — no HTML, CSS, or JavaScript required. The Go process handles state sync and WebSocket brokering; your Python app just pushes card updates to it. It's aimed at data scientists and ML engineers who want interactive UIs without frontend skills.
The Go server architecture is genuinely clever — because state lives server-side and gets pushed to browsers over WebSocket, you can update a live dashboard from any async background task without wrestling with threading or SSE boilerplate. The 250+ working examples cover almost every component type, which makes it fast to prototype. R support is rare in this space — Shiny is the only real competition there, and Wave's layout system is more flexible. The FedRAMP compliance angle is a real differentiator for anyone building internal government or enterprise tooling.
The 'no frontend knowledge required' pitch breaks the moment you need anything outside the 80 built-in components — custom components require TypeScript and a build step, which defeats the whole pitch for most users. Running the app requires the Wave binary as a sidecar process alongside your Python script; deployment is messier than Streamlit or Gradio where it's just a single pip install. Community traction is weak — 4000 stars after five years puts it well behind Streamlit, Gradio, and even Dash, which means fewer Stack Overflow answers and a thinner ecosystem of community extensions. The tight coupling to H2O's own cloud platform shows through in the docs and examples in ways that make the open-source self-hosted path feel like a second-class citizen.