Simplifying an open source machine learning app
Create an open source community app that teaches and supports machine learning testing
THE CHALLENGE
An IBM student led project needed a team to take the early concept to production ready. The app aims to educate people with no experience with machine learning, and provide an open source platform for data scientists and system developers to find, upload, and compare machine learning models.
How I helped
To gain a deeper understanding of the machine learning domain and its current competitors, I conducted a comparative analysis. This helped the team better comprehend the landscape data scientists and system developers inhibit, and bring context to how they might utilize this tool in their research.
Comparative analysis
The initial version was functional but not user-friendly and did not align with the organization's overall objectives for a production-ready product. To address these issues, I collaborated with the team on several object mapping exercises to fully understand the system's scope and document the intricate relationships between objects.
Object mapping workshop
The business objectives were to support both free and authenticated users. I guided the team in various role and permission workshops, and assisted in documenting the requirements.
User role requirements
Using a navigation flow diagram of the system objects, I led the team of developers and stakeholders through various discussions and identified ways to phase down the approach for the MVP.
Strategy Alignment
Conducted monthly usability studies on new features. This included creating research plans with the team, and analyzing the observations to present areas of improvements based on severity of the problems seen.
Usability studies
Requirement improvement workshops
I led team wide workshops to identify ways to iterate and improve on the problems seen in usability testing.
Throughout the development process, I developed various sets of wireframe concepts that detailed the complex workflow and interactions to find, test, and compare machine learning models