Hello! Yes, we are working to get there. Recently I was able to bring our Kotlin build up to TF 2.3.X compatibility. We are integrating now for Kotlin JVM with the Java TF Special Interest Group. There is still work in progress on this, but we're getting there.
Regarding tensorflow recommenders, the Kotlin library should eventually have support for this, and more. Part of the Kotlin project's goal is to extend the typical Keras libraries with an additional DSL to support higher-level AI/ML programming, so developers can focus less on the mechanics of model configuration, and more on building applications with ML capabilities, with less modeling tedium.
I should have an article up in the coming weeks on
https://tfk.dev explaining more of the planned features and capabilities for TF Kotlin 1.0 release.