Case Study: iFood achieves sub-millisecond ML performance with Redis Cloud

iFood relies on Redis Enterprise Cloud as foundation for ML operations

iFood, one of Latin America’s largest food delivery platforms, needed to support real-time machine learning operations as order volumes and user demand surged. The company’s challenge was to process feature data quickly, keep ML logic consistent across teams, and control infrastructure costs while running more than 100 production models. To address this, iFood used Redis Cloud on AWS and Redis on Flash as the foundation of its ML feature store.

Redis helped iFood build a high-performance, scalable feature store with sub-millisecond read performance and lower storage costs by tiering hot data in DRAM and colder data on Flash. The solution improved collaboration and reusability for ML engineers, supported fast microservices interactions, and strengthened reliability and fault tolerance. According to iFood, Redis Cloud also reduced infrastructure management overhead and saved money, while helping the team keep latency under 200 milliseconds for user interactions.

iFood

Daniel Galinkin

Head of Machine Learning Engineering