Story
Confluent
CMS EnergyCMS Energy Uses Flink on Confluent for Feature Pipelines
CMS Energy, a utilities organization in the United States, uses Flink on Confluent from Confluent to support applied machine learning for data scientists and ML engineers.
Value results
| Category | Value result |
|---|---|
| Productivity | Fewer stalled items because feature pipelines has a clear owner |
| Productivity | Handoffs in applied machine learning sit in a shared queue instead of a mailbox trail |
| Capability | New joiners can see how applied machine learning actually runs |
Story
Utilities work at CMS Energy spans more than one site, even when headquarters sits in the United States. Feature pipelines was splitting across regional habits. Data scientists and ML engineers asked for a shared way to run applied machine learning without freezing local judgment.
Confluent (Flink on Confluent) is what they standardized on. Confluent is the data streaming platform built around Apache Kafka, used to move events between applications in real time. CMS Energy uses it as the system of record for feature pipelines, with data scientists and ML engineers as the primary operators and other groups coming in through the same queue.
Leaders get a picture they can actually walk. Teams get fewer mystery statuses. The story is about operating change, not an unpublished percentage.
