h2oai/driverlessai-recipes ? reverse-engineered prompt

Build me a Python repository of sample custom recipes for H2O Driverless AI.

I want it organized as a collection of reusable examples for transformers, models, scorers, explainers, data prep, and a few utility scripts for loading recipes in air gapped setups. Include clear README guidance for how to write a recipe, best practices, and a FAQ style section about safety and performance. Make the examples feel realistic, like things people would actually upload into Driverless AI to extend feature engineering and model building.

Please add a good mix of simple and advanced recipes, with comments so someone can learn from them. It should be easy to browse and use as a reference library, not just a few random scripts. If you think it helps, organize the examples into sensible folders and make sure the repo has enough structure that someone could expand it with more recipes later.

Use plain Python and keep it compatible with how Driverless AI custom recipes are expected to run. Look up the current Driverless AI recipe docs online if you need to.