Multi agent orchestration for Julia code generation: Best tooling / practices?

Hello all,

I’m hoping to crowd source some information on best tooling and practices for multi-agent orchestration for production of Julia code. I am starting to get on the bandwagon of offloading relatively straitforward but time consuming coding tasks to multi-agent systems (though I am by no means a power user). This is all fine and good when your writing Python code since you can just orchestrate within Python. Unfortunately the Julia bug bit me and I’ve transitioned most of my work to Julia over the past 4 years. Any thoughts on best approach to this for Julia code generation / testing? I have a couple of approaches in mind but I would rather hear from those of you that are far more well versed in this than me.

To be clear, this is a tooling question, not a question of how to get agentic-orchestration to be effective.

I did find this thread (AI tools to write (Julia) code (best/worse experience), e.g. ChatGPT, GPT 3.5 - #61 by findmyway) but it is almost 18 months old at this point.

Apologies if this is a basic question and I’m 6 months behind the times.

Not quite sure what exactly you are asking. Generally AI works similar for Julia as it works for other languages I think. Here are two posts about using AI coding agents for Julia that you maybe find interesting:

  • I had some success using Roo Code (VSCode plugin) for translating some Rust package to Julia
  • Chris Rackaukas explains a bit how he herds an army of Claudes:

Does either of that go in the direction of your question or did I miss the point entirely? :sweat_smile: