using-ai-engineering

Routing directs your AI/ML task to the correct Yzmir pack based on problem type—whether you're training models, deploying, fine-tuning LLMs, or working with reinforcement learning. Describe your task and the Skill matches it to the appropriate pack, loading only necessary components.

Use pack selection when starting an AI/ML project or unsure which pack applies. It handles training, deployment, RL, fine-tuning LLMs, and multi-domain tasks by enforcing mandatory clarifications and ensuring problem-type matching before routing.

Yes, the portable workflow supports combining multiple domains—for example, training an RL model then deploying it—by routing tasks sequentially to each appropriate pack based on your problem type at each stage.

Mandatory clarifications stop ambiguity by asking targeted questions when terms are unclear, ensuring you reach the correct pack without wasting time on trial-and-error with the wrong tools.

Multi-domain tasks are routed sequentially—each component (training, RL, LLM fine-tuning, deployment) routes to its corresponding pack independently, so you can chain packs together in one workflow.