Covid-19 Dashboard

The Story Behind It

I had just completed my MSc in Data Science at Birkbeck, University of London, and was in the middle of a job search when Adam Halasz — a mathematics professor at West Virginia University and one of the most formative people in my academic life — reached out. He wanted a dashboard that used the Johns Hopkins dataset rather than the numbers the state was reporting, to get a clearer, more reliable picture of how the pandemic was actually developing across the US. Could I build it?

It was exactly the right project at exactly the right moment. I'd spent a year learning Dash, Plotly, cloud deployment, and data pipelines — and here was a real problem, for a real person I respected, that I could actually solve with those tools. Adam had a specific request for the R-rate estimation, and I used the EpiEstim package in R to implement it properly — a Bayesian epidemiological method that produces estimates with uncertainty bounds rather than a crude rolling average.

The dashboard was deployed on Google Cloud Platform and ran as a live public resource. In the end it was used primarily by Adam himself, and eventually shut down. An attempt to host it through the WVU Mathematics department didn't survive the politics of institutional websites. But that wasn't really the point.

Why Adam Matters

During my undergraduate years at WVU I was genuinely lost — moving between psychology, biology, physics, philosophy, mathematics, and statistics, drawn to all of it and unsure how any of it connected. Adam was the person who helped me find the thread. Working as a research assistant in his lab, analysing datasets generated by flow cytometry machines, I saw for the first time how mathematics could be applied in a real research setting — how the abstract became useful, how data told stories about biological systems.

That experience is what eventually led me to pursue an MSc in Data Science. Building this dashboard was a way of completing a circle: taking everything I'd learned in the years since, and giving something of genuine value back to the person who first pointed me in the right direction.

Technical Details

• Data

  Johns Hopkins CSSE — Daily case counts sourced from the Johns Hopkins University Center for Systems Science and Engineering public repository, chosen specifically for its reliability over state-reported figures. Parsed and cleaned with Pandas.

• R-Rate

  EpiEstim (R package) — The reproduction number R was estimated using the EpiEstim package — a Bayesian epidemiological method that uses the serial interval distribution of Covid-19 to produce robust estimates with uncertainty bounds. Included at Adam's specific request.

• Visualisation

  Plotly Dash — Interactive choropleth maps, time series charts, and county-level drilldowns built with Plotly and served via the Dash framework — a Python-native way to build analytical web applications without writing JavaScript.

• Deployment

  Google Cloud Platform — One of my first public deployments built for someone else. Containerised and hosted on GCP, giving me hands-on experience running a live, data-driven web application for a real audience.

What It Represents

This is a smaller project technically than most of what I build now. But it sits at a meaningful junction in my story — the bridge between an undergraduate spent wandering across disciplines and a career spent applying all of them. The skills were new, the tools were fresh out of an MSc, and the person it was built for was the one who first showed me why those skills were worth acquiring.