Outcome forecasting
Probabilistic models that turn live coursework into grade and GPA forecasts, updated as every assignment lands.
Pyko Labs builds predictive models of human learning and deploys them at scale in Pyko, toward a general theory of how minds improve.
Education technology has spent two decades recording what students did. We model where they are going.
The lab was founded inside Pyko to answer the questions the product kept raising. Can an outcome be forecast honestly? Can software sense when attention is about to break? Can a system understand a course the way a student has to? We publish into the product first: a model is not finished until it has survived a semester.
RESEARCH DIRECTIONS
What we work on
Attention and cognitive rhythm
How focus behaves across a day and a term: session pacing, energy cycles, and when effort becomes retention.
Knowledge representation
Concept graphs that map what a course actually asks of a student, and where understanding breaks down.
Anticipatory interfaces
Ambient systems that surface the right thing before it is asked for, and step back when they are not needed.
The lab behind a living product.
Pyko Labs is the research group inside Pyko, the study platform built at the University of Toronto. Every model in the product started here, and every question in the lab started with a student. That loop is the method: the semester is our benchmark, and shipping is part of the experiment.
Technology should not stand between you and your focus. It should flow with you.
The Pyko design philosophy
Help us model how people learn.
16 open roles across research, modeling, applied AI, product, data, and safety. All based in Toronto, hybrid or remote.