Hi, I would like to learn how the libraries within the “Kotlin for data science” ecosystem work ( https://github.com/Kotlin/kotlin-jupyter#list-of-supported-libraries ). Are they written entirely in Kotlin , or is it a bridge to the Python environment? Does “Kotlin for data science” require a Python environment installed on the system? What about libraries that are written in native C/C++ in Python, such as numpy?

Kotlin has no interop with python as known, for now we have libraries written totally in Kotlin or with kotlin-java-jni-c api - c++ bridge No needs to install Python for KotlinDL, Dataframe, Kandy

Sure, it could be truth only for kotlin-numpy, but it's deprecated now https://github.com/Kotlin/kotlin-numpy Also, "supported" means that we tested the library in some way with notebooks and maybe provide some integration features such as renderers or default imports for it. But actually every library that has JVM support is supported.

So what about kotlin-jupyter-kernel ? Is it kotlin or python solution ? Is the notebook code compiled to bytecode jvm or interpreted by python ?

Kotlin notebook kernel itself is pure JVM. If you use Jupyter, its backend is written in Python. IDEA notebooks are pure JVM.

Thank you very much for explanation, rewriting these libraries must be a big effort. Great job!