AI Data Lake · Huawei
When the data is invisible,
the anxiety is not.
The Bet
Enterprise data tools default to showing everything — every node, every pipeline, every anomaly. For data scientists navigating hospital records or financial datasets, more visibility doesn't reduce confusion. It compounds it. I bet on a different principle: don't simplify complexity, make it navigable. Show the right thing, at the right moment, to the right person.
The Work
Three decisions defined this project.
Nodes that think in layers
Data scientists tracing corrupted data need to read relationships at different depths simultaneously. I designed a node card system with rich, differentiated states, each card surfacing the right density of information for its role in the lineage path. Paired with an intelligent side panel that responds in visual sync, the system creates a felt sense of path and direction. Users don't read the lineage. They feel it.
A sky of data points
Assessing dataset quality across hundreds of entries needed a different kind of interface. I designed a star-ring chart: datasets distributed as points, with data quality scores on a draggable horizontal axis. Color coding makes dataset health immediately readable at a glance.
When the visual language is the aha moment
The entire product is organized around a single concept: data as stars. Charts, panels, and interaction patterns all extend from this visual metaphor. The result isn't just consistency, it's memorability. The interface has a point of view. That's rare in enterprise software, and it's what makes it stick.
The Impact
Confidential, full case not available right now. The visual language was adopted internally across the storage product line.
Role
Product Designer
Type
Systems Design · AI Infra