Apache NiFi and Apache Kafka are two different tools with different usecases that may slightly overlap. Here is my understanding of the purpose of the two projects.
NiFi is "An easy to use, powerful, and reliable system to process and distribute data."
It is a visual tool (with a REST api) that implements flow-based programming to enable the user to craft flows that will take data from a large variety of different sources, perform enrichment, routing, etc on the data as it's being processed, and output the result to a large variety of destinations. During this process it captures metadata (provenance) on what has happened to each piece of data (FlowFile) as it made its way through the Flow for audit logging and troubleshooting purposes.
"Apache Kafka is publish-subscribe messaging rethought as a distributed commit log"
It is a distributed implementation of the publish-subscribe pattern that allows developers to connect programs to each other in different languages and across a large number of machines. It is more of a building block for distributed computing than it is an all-in-one solution for processing data.
