QuestionQ177

Custom Model Serving

A machine learning engineering team has determined that predictions must be available for querying in continuous, equal-sized increments. A computation may be included in an increment once all its feature values are present in the inference Spark DataFrame.

Which of the following tools can provide this form of continuous inference?

  • A MLflow
  • B Delta Lake
  • C Spark UDFs
  • D Structured Streaming
Explanation

Apache Spark Structured Streaming executes processing incrementally as data arrives, commonly in micro-batches. It can therefore score records continuously as soon as the inference DataFrame contains the feature values required for a computation.

Learn more

Community Discussion

No comments yet. Be the first to start the discussion!