Improving DX through UX Processes
I am working on a project called Harp which is a HPC-Cloud convergence framework for Big Data. It is basically a library which runs complex datasets converging with HPC which can be used for Machine learning projects.
The Goal
To make Harp’s library more usable and developer-friendly.
The Approach
Understand how data scientists and developers work
- Understanding what they do by having co-design sessions
- Understanding the product by attending data workshops
- Understanding how they use and implement the code by having co-design sessions
Improve Documentation UX by
- Restructuring Information Architecture
- Redesigning vector diagrams that explain different algorithms
- Restructuring code snippets
Improve Website experience
- Include search and micro-interactions like copying to clipboard button
- Download packaging
- Version and Release notes
Packaging APIs
Phase 1: Co-design sessions with Data scientists + Data workshops
I personally conducted a workshop with researchers at IU’s Intelligent Systems to figure out the structure of the documentation and which component needs to go where. I also did user interviews to see what data scientists and developers care about while using a library.
A HPC data workshop conducted by my team in which data scientists from universities and orgs like Intel, IBM and NEC were invited.
Coming up with DX factors
Improving the Information Architecture
Actual Implementation
In-progress (version 2.0)
Learning and future work
Developer Experience is a very exciting area to explore as it has a different set of challenges and expectations than Consumer UX. Harp will continue to grow and there are multiple sub-projects which are in the pipeline. I will keep updating the progress here as it develops.
Here are the links to all of the projects-
