
A fact-checking platform for Goodly Labs. Volunteers annotate news articles for reasoning and evidence problems, their overlapping judgements get reconciled into a single credibility report, and readers see the result layered over the original article. I worked across the stack: the Django models and workflows behind annotation, training data, and scoring, which now back over 2,000 published reports; the reader’s annotation engine, where several claims can highlight the same sentence without any of them losing their own context; and most of the browser extension people use to submit an article without leaving the page they’re on.