Case Study: Gojek scales ML operations and cuts costs with Union.ai Flyte

Gojek - Customer Case Study

Gojek, one of Southeast Asia’s largest multi-service and digital payment platforms, needed a scalable, Kubernetes-native orchestration platform for its growing machine learning workloads. Its legacy ML pipeline setup, built on top of Clockwork and Airflow, had more than 1,000 DAGs but was difficult to test locally, had a steep learning curve for data scientists, and was facing scheduler scaling and infrastructure overhead issues.

Gojek chose Union.ai’s Flyte platform to modernize ML orchestration with workflow versioning, Kubernetes-native execution, and a powerful Python SDK. With Union.ai’s Flyte, Gojek expects to onboard 100–200 new workflows per month, is on track to deprecate its legacy platform within six months, and has already seen 20%–80% cost savings across migrated pipelines.

Gojek

‍Pradithya Aria Pura

Data Science Platform Team