Case study
COVID-19
A COVID-19 dashboard and the data parser behind it, built to make the numbers legible to the public, and run publicly for around eighteen months.
Overview
A public COVID-19 dashboard and the parser behind it: collecting case data, keeping it current, and presenting it so that a non-specialist could read a trend without being misled by it.
Built alongside Muhammad Umer Farooq, and run publicly for around eighteen months.
What I own
The parsing and data pipeline, with front-end work alongside.
Complexities tackled
The source data was neither clean nor stable. Reporting formats changed, regions were renamed and figures were revised after publication. A parser that assumes today’s shape breaks silently the first time a column moves, so the collection had to validate what it received rather than trust it.
Speed was a requirement, not a nice-to-have. A dashboard people check daily has to load immediately, which means the expensive work happens at collection time and the page only ever reads a prepared result.
Presenting figures honestly is a design problem. Cumulative counts always rise and always look alarming; daily counts are noisy and always look encouraging on a good day. Choosing what to show, and over what window, decides what the reader takes away, and getting that wrong on public health data matters more than usual.
Stack
Python for the parser, Vue with TypeScript on the front, Bootstrap for layout.
Working on something similar?
Tell me what you are building and what is in the way. I will tell you honestly whether I am the right person for it.