AI-Powered Event Recommendation Platform for Personalised Experiences

The challenge

The organisation operated a large-scale digital sports marketplace serving over a million users and thousands of events annually, but lacked an effective recommendation capability. User interaction data such as searches, clicks, and views was not systematically captured or centralised, limiting the ability to deliver behaviour-driven recommendations. Data was fragmented across operational systems and not structured for analytics or AI use cases. Existing recommendation experiments were not production-ready, and there was no scalable or governed architecture to support personalisation. Additional constraints around cost efficiency, limited in-house data science capacity, and GDPR compliance increased the complexity of delivering a robust AI-driven solution.

The approach

Designed and validated a cloud-native recommendation architecture on Microsoft Azure, focused on combining interaction data with operational event data. A dedicated analytics layer was introduced using Azure SQL to model event-interaction relationships, supported by ingestion and transformation pipelines through Azure Data Factory and event-stream processing. Azure Machine Learning was used to build an initial ranking model alongside heuristic scoring to generate ranked event recommendations. Azure OpenAI capabilities were incorporated to support enrichment and future explainability use cases. The solution also included a containerised recommendation API, CI/CD deployment workflows, secure secret management, and an architecture aligned to Azure Well-Architected principles for scalability, security, reliability, and operational governance.

The outcome

Delivered a validated blueprint for an AI-powered recommendation platform that improves event discoverability and personalisation across a digital sports environment. The solution established a scalable and cost-efficient foundation for using real user behaviour to drive more relevant recommendations and improve engagement. It also created a governed architecture for future expansion across web, email, and mobile channels, while improving data readiness, operational visibility, and the organisation’s ability to evolve its AI-driven personalisation strategy.