Pushing
Boundaries.
We build representations of people and intent that go beyond surface-level text matching, enabling precision compatibility at scale.
Matching supply and demand underpins the global economy. Jean aims to power every match.
Technical Writing
Deep dives on matching infrastructure, embedding models, and the technical decisions behind our platform.
View All Posts
Research
Research Areas
Person Embeddings
Modeling complex human traits, intent, and historical behavior to map compatibility across disjoint domains.
Domain Outcomes
Training dual-encoder models on successful, real-world outcomes rather than text similarity.
Behavioral Context
Capturing the longitudinal signals that turn static profiles into living representations precise enough to match on.
Techniques
REPRESENTATION LEARNING
CONTRASTIVE LEARNING
TRANSFER LEARNING
SPARSE AUTOENCODERS
HYPERBOLIC EMBEDDINGS