/* case study */
Analytics Web Services Layer
Championed and architected a Go web services layer for a big-data visualization platform built on a GPU-accelerated RDBMS.
// impact: Laid the foundation for far more robust platform features
// Context
HEAVY.AI's analytics platform let users visualize huge datasets on a GPU RDBMS, but the client application talked to the database layer directly.
// Challenges
- ›No dedicated web services layer between the UI and the GPU database
- ›Large areas of undocumented JavaScript, some more than five years old, written by engineers who had since left
// Approach
- ›Designed and built a Go/Echo web services layer that integrated with the Apache Thrift API and exposed REST to the client
- ›Re-architected the Node.js application layer with Koa and react-router, modularized for future serverless distribution
- ›Added full-duplex communication with Socket.io
// Outcomes
- ›Opened a path to far more robust and complete platform features
- ›Shipped data-viz demos such as a COVID-19 dashboard aggregating WHO, CDC and Johns Hopkins data