Launchpad

Overview

The output of data scientists sometimes lies in the form of interactive R or Python applications that are built for users within other parts of their organization. Traditionally, data scientists would create bespoke applications and then share these out with colleagues in emails or presentations. This process lead to duplication of work by data scientists, unclear metrics of success, and confusion from non technical users on how to use the application. Data scientists needed a way to create and manage these applications and share reports of usage with management. The users of the apps (their colleagues) wanted the apps to be more discoverable, learn how to use the app, and give feedback to the creator.

My Role

  • Led design sprint with designers, engineers, and PMs to understand project and constraints

  • Conducted usability testing and user research with internal stakeholders and customers

  • Created all wireframes, produced all high fidelity mockups for this feature 

  • Worked with frontend engineering to ensure clean buildout of UI components, interactions, and user flows

Project Goals 

Publishing and Monitoring (For Data Scientists)

  • Ability to publish R Shiny or Flask Python Apps, APIs, Web forms, and reports. 

  • Show metrics on app views - this is often a metric of success within data science organizations
  • Bring UX clarity to app workflow: creation, permissions, status, and usage.

Launchpad (For non technical users of data science apps)

  • Provide a way for non technical users to view without drilling into specific projects and code
  • Ability to learn about the app and provide questions or feedback to the creator
 
 
 

Publishing and monitoring an app

Wireframes

 

V1 Mockups and Process

 

v2 Mockups - Establish UI componentization and improve UX flow

 
 

Launchpad

Wireframes

 

Mockups

 

V1 Mockups and Implementation

 
 

Outcomes and next steps

 

The launchpad received huge success and validation before a line of code was written. We maintained a constant feedback loop by talking to internal stakeholders and customers, testing paper wireframes and interactive prototypes, showing a prototype at Rev Conference, and launching to a beta set of customers. Launchpad is currently launched in beta and is being rolled out to cloud and on-prem customers. The current version only showcases applications (Shiny and Flask applications). The next steps are to integrate API and reports, and also improve the experience of filtering through a large subset of data science products.