Show Me Your MVP!
summary
Claim: data scientists can own product demos without front-end work by building cloud‑hosted interactive prototypes that stakeholders can exercise and annotate. Practical build: a proof‑of‑concept for an internal opportunity marketplace that stitches dozens of models into a pipeline, implemented with Streamlit, using caching and session_state, an ag‑grid table for selection, and buttons that save feedback as JSON to a logfile. Deployment runs on a cloud VM under tmux; the demo surfaces edge cases and secures early buy‑in.
AI-written summary · about
chapters (20)
- 0:00Introduction
- 0:27Audience MVP experience
- 1:33Introduction
- 2:58Project motivation
- 4:41Defining MVP
- 5:23Data scientist roles
- 7:12UI for data science
- 8:41UI criteria
- 10:26Low-code Python frameworks
- 11:29Streamlit for MVP
- 13:13Streamlit book app demo
- 14:16Streamlit caching
- 16:07Interactive UI components
- 17:56Session state
- 18:17User input
- 19:01Interactive table widget
- 20:28User feedback
- 23:22Deploying Streamlit
- 24:28Summary
- 25:55Q&A
related talks
42mStreamlit: A Faster Way to Build and Share Data Apps
05mDemo This!
01h 58mHow to create beautiful interactive GUIs and web apps
21mWeb apps for data science using streamlit
46mLightning Talks
58mCreating Interactive, Animated Reports in Streamlit with Vizzu
18mRapid Prototyping in AI Project using Colab and Streamlit
28mBeyond Prototyping: Building Production-Level Apps with Streamlit