Rambler&Co — Search and Recommendations

Rambler's digital media ecosystem had more than 45 million monthly active users across its properties. Search and recommendations were the systems that decided what those users saw and what the ecosystem earned from it.

This was a media ecosystem context: search and recommendations sat at the intersection of user intent, content distribution, advertising economics and machine-learning quality.

Search monetization and recommendation quality were limiting both revenue and engagement. Improving the ranking model alone would not have been enough, because the surfaces, incentives and quality metrics around it determined what the model was allowed to optimise for.

I led the Search Technologies and Recommendation Services direction, with approximately 10 direct team members and broader cross-functional involvement across the ecosystem.

Treat ranking quality, distribution surfaces and monetization as one product system rather than optimising the model in isolation.

Measure recommendation success on engagement depth as well as click-through, so the system was not rewarded for shallow curiosity.

Improved search monetization and SERP performance.

Built and led personalized recommendation products across the ecosystem.

Connected product analytics, ranking quality and monetization decisions.

Recommendation systems are product systems, not just models. Distribution surfaces, incentives, quality metrics and editorial context all change the outcome.