SWIM – Scalable Workloads for Iterative Machine Learning
Company: Lowe’s
Role: Senior Product Designer (sole designer)
Timeline: 2023–2024
Tools: Figma, Miro
Project Overview
SWIM is an internal Lowe’s platform designed to make machine learning accessible to non-technical users.
Business analysts used SWIM to perform sales forecasting, assortment planning, and marketing analysis—all without coding.
It brought together backend ML capabilities into a unified, intuitive interface.
The Challenge
The goal was to simplify complex ML workflows for analysts unfamiliar with model configuration or data science. Challenges included lack of a centralized UI, high learning curve, and unclear workflows.
The solution needed to provide structure and clarity while staying scalable.
My Role
As the sole Senior Product Designer, I led the UX strategy and execution. I collaborated with PMs, ML/AI strategists, and data scientists to translate technical processes into approachable interfaces.
​
-
End-to-end UX design and task flow creation
-
Wireframes, prototypes, and final UI design
-
WCAG 2.1 AA-level accessibility
-
Internal demos and design validation
The Process
Stakeholder interviews with Lowe’s AI/ML Center of Excellence revealed that analysts needed a non-technical way to use machine learning. Pain points included disconnected tools, complexity, and unclear outputs.
I co-created a task flow with data science SMEs. Using this as a foundation, I built wireframes, high-fidelity mockups, and interactive prototypes. Accessibility was baked in from the start.
Initial feedback came via internal design shares and demos. Usability testing with analysts was planned for post-MVP launch.
The Solution
-
Home Page: Dashboard displaying real-time status of jobs and pipelines, with clear CTAs for creation.
-
Job Creation: Dataset upload, field mapping, and model association, with simple guided flows.
-
Pipelines: Visual editor to sequence jobs and schedule autonomous execution.
Results & Impact
The product launched successfully with positive feedback from analysts and stakeholders. Analysts were able to independently access ML-driven insights and integrate them into business planning.
Reflection
I gained foundational understanding of ML and learned the value of collaborative task flow design. In the future, I’d push for more upfront whiteboarding with data scientists and expanded user research to guide design decisions earlier in the process.
