Booking.com Evolves AI Infrastructure and Platforms
At QCon London 2026, Jabez Eliezer Manuel, Senior Principal Engineer at Booking.com, outlined the company's 20-year AI evolution, detailing data management, machine-learning engineering, and domain intelligence efforts. He described a seven-year migration off Hadoop, a current inference platform serving over 480 models and 400 billion predictions per day with sub-20ms latency, and four unified domain platforms—GenAI, Content Intelligence, Recommendations, and Ranking—for personalization.
Key Points
- 1Describes migration off Hadoop after scaling limits, moving to NVMe-backed smaller DBs and distributed systems
- 2Highlights production scale: 480+ models, 400 billion daily predictions, sub-20ms latency demonstrating real-time capability
- 3Encourages unified domain platforms and interleaved experimentation to accelerate safe personalization and efficient model deployment
Scoring Rationale
Practical, high-credibility engineering lessons from Booking.com's migration; limited novelty and moderate generalizability outside travel sector.
Sources
Public references used for this report.
Practice interview problems based on real data
1,625 SQL & Python problems across 15 industry datasets — the exact type of data you work with.
Try 250 free problems
