Python, Rust, C++, Golang, and Swift come to mind for me, depending on the project, but I'm starting to think Java is actually the ideal LLM language, especially with LLM-driven sysadmin-ing making it easier to set up and run Java.
I've also authored a Machine Learning Model compiler for the JVM: https://github.com/exabrial/petrify . This allows you to take Many Trees/Ensembles/Regressors/Classifiers/etc and compile them as regular Java classes! ...removing the long dependency chain that ML Models drag with them to run in production. It's also very memory efficient; exactly one array allocation invocation, with 0 other Garbage Collector pressure. Your weights and splits are also stored in the constant pool; making it an excellent C2 target and do not add to the additional heap memory pressure!
Overall, I think it's an ideal language for Data Science; but what it lacks compared to Python is community engagement. Oracle has Tribuo, which is actually quite good; but Oracle is still learning to interact with the open source community.
In my own Python code, after tightening up Pydantic + strict mypy + pytest, I’ve noticed a significant increase in Claude’s first-pass accuracy. I believe which conventions you adopt within a language is more important than the language itself.
Additionally, a pitfall specific to Java is that the training data still contains a large amount of Spring XML/AbstractBeanFactoryAware-style patterns from the old Java 7–8 era. To elicit modern Java 21+ idioms, you need to specify them explicitly in the prompt. Conversely, Python tends to produce the latest idioms by default.
I understand the OP’s intuition (that Java is being reevaluated due to improvements in infrastructure tools), but the essence lies in “language design that enforces LLM-friendly coding styles.” In that sense, my personal conclusion is that Java is one of the good fits.
Why?
They write python as readily as people breathe, though.