UofT - Embedding the Canadian Job Market: Building a Job Discovery Engine

This demo presents how we built a multi-interface job discovery platform organized around a shared AI and analytics pipeline. The system supports three roles: a student interface for discovering opportunities, an employer interface for creating structured job postings, and an admin interface for monitoring platform analytics and data quality. The technical focus is a content-based recommendation system built using sentence-transformer embeddings and tag-based metadata. Jobs are embedded during AI assisted ingestion, and user preference vectors are constructed dynamically from saved and recently viewed jobs. Recommendations are generated by computing cosine similarity between the user embedding and active job embeddings. We will walk through the architecture of the pipeline and discuss how structured job templates, behavioral signals, and analytics instrumentation helped improve the quality of recommendations and observability of the system.