HN user
The main difference is the layer where diskless is introduced. KIP-1150 makes it a diskless topic path with a Batch Coordinator; AutoMQ tries to keep the existing Kafka broker/leader path and swap out the storage underneath.
That makes the designs feel quite different even though both end up using object/shared storage.
Seems like a smart tool for enhancing coding skills. Tagging LeetCode problems by difficulty and having a tailored review plan based on spaced repetition is a strategic approach to learning.
What are the side effects of AI that you mentioned?
I need such a tool. The only concern is whether this advisor can really give reliable indexing advice? Is it based on cost or rules? Maybe I would dare to use it after knowing more about its implementation.