Data science and information design
Pontevedra, Galicia, Spain
alexcorvin.io · Work · LinkedIn · hello@alexcorvin.io
I'm a data scientist, and I work with the data that organizations already have but rarely look at: spreadsheets, records and research that exists only as PDFs. A 2020 IDC survey for Seagate found that 68% of the data available to businesses goes unused.
Every project starts with that raw material. The records come first, then the metrics, then a single question, and only after that the design. The method and the limits of the data stay on the page beside the chart.
Spain Housing Prices. House prices rose in all 17 of Spain's autonomous communities in the year to the second quarter of 2026, from 9.5% in Navarra to 15.0% in Asturias. The map shows the change for all homes, new homes and second-hand homes, and a line chart follows the national rate back to 2016.
Case study · Live page · Code
AI Data Center Power. Ten data centers hold 39% of the AI power capacity that Epoch AI tracks. The page draws the 73 sites with a power figure as circles sized by megawatts, ranks the ten largest by power and by computing, and opens each one to show its owner and users.
Case study · Live page · Code
Contaminants in Cat Food. The Clean Label Project screened 100 cat foods from 62 brands for more than 122 contaminants, but it hasn't released results for individual products. This study works only from the figures the report published, and it gives a panel to what those figures can't show.
Case study · Live page · Code
Earlier studies apply the same method over longer stretches of time. The Speed of News measures how long news of eleven events took to reach American readers between 1799 and 2023, and The Real Movers compares renewable energy shares across countries from 2014 to 2024. There's also a Medicaid spending map, and Holding Still, which measures how rare slow-shutter photographs are and includes my own.
My work also includes private document search that runs on local models and answers questions with a pointer back to the source passage, plus Python pipelines that take over repetitive record work. The tools are Python, Svelte, D3 and Ollama.
To start a conversation, write to hello@alexcorvin.io.