Stratum

PROBLEM
Organizations across industries, from finance to healthcare to retail, sit on enormous volumes of raw tabular data with no fast, accessible way to extract intelligence from it. Building custom ML infrastructure is expensive, slow, and requires specialized talent most teams don't have internally.

ROLE
As designer and builder, I conceived, designed, and developed the entire frontend of Stratum a data intelligence platform that lets any data-driven organization connect datasets, train AI models on their own data, and query insights in natural language. I worked in close collaboration with backend developers to define the data architecture and API contracts, using AI-assisted development to accelerate the build from concept to functional prototype.

PROCESS
I designed the full product experience across five core surfaces: a dashboard with model and dataset oversight, a data browser with filtering and schema preview, a model training interface with live progress tracking, an AI playground with pre-built and live conversations, and an analytics layer with automated insights and usage metrics. Every design decision prioritized clarity and trust making complex ML workflows feel approachable without oversimplifying them.

OUTCOME
A fully functional, browser-ready SaaS prototype demonstrating a complete end-to-end data intelligence workflow, from raw dataset connection to deployed model endpoint. The project established a scalable design system and interaction language applicable across verticals, and proved that a single designer working with AI tooling can compress what would traditionally be a multi-month product sprint into a focused, high-fidelity deliverable ready for engineering handoff.