Creating high-performing, widely applicable ML models does not always require the same data processing approach. Use cases at Intuit derive from two main approaches — batch and streaming. Batch processing is scheduled and deals with data collected in advance. On the other hand, stream processing is continuous and real-time, meaning that the data will go through an event store like Kafka, before going through a feature processor.
In this webcast, we’re joined by Juhi Dhingra and Dunja Panic. Working as product managers for Intuit’s batch and streaming platforms, they will discuss the process of creating these two platforms, walk us through the real business use cases where stream and batch processing is used for feature generation and how they are helping their customers decide which platform will be most suitable for their needs. They will also discuss their plans to create unified tooling for batch and streaming feature generation.