Recommender systems at Seznam scale
Discovery and ranking across 50M+ requests a day: a balanced history-based recommender, clickbait and quality classifiers, agentic content crawlers, and the experimentation framework to prove each change.
Work
Client engagements are named or anonymised with permission. The rest is prior in-house work, reframed around what the system did and what changed.
Discovery and ranking across 50M+ requests a day: a balanced history-based recommender, clickbait and quality classifiers, agentic content crawlers, and the experimentation framework to prove each change.
An email platform that began as a large e-commerce retailer's internal sending tool in 2009 and grew into a SaaS running 100M+ emails a month for 196 e-commerce brands across five countries.
Product strategy for a DMP processing 500M+ user events a day for 50+ media agencies: real-time ingestion, self-service segmentation, and a privacy-first (GDPR) data model.