This note is for practical engineering notes on working with AI. For conceptual and philosophical notes see AI.
Dev setup
Notes on how I work with with AI tools like Claude Code is in AI dev tooling for the moment. It's missing a lot of details about git worktrees and running multiple parallel agents.
Python
I have a note on Python.
I have a note on pytorch. I have a note on pandas. I will mostly use python when exploring models on Hugging Face.
I have a note on Jupyter Notebooks.
Elixir
I have a note on Elixir. I think Bumblebee was an important library for Elixir AI but haven't explored it as much as python tooling.
Elixir LiveBook is an Elixir approach to something like Jupyter Notebooks and has a way to run backend locally. You can then easily create "smart cells" that will do a bunch of the heavy lifting for you.
Lean
I have not really explored Lean apart from their opening tutorial but I feel like it has fascinating implications and I'm very interested in further explorations in formal verification, particularly as a more rigorous extensions of testing and spec driven development.
Other
AI Ops, as in, how to run AI workflows for users of an AI service, is an area I have not explored, but I have collected some links for future exploration. I know nothing about this at the moment.
- https://dspy.ai/
- https://mlflow.org/
- https://phoenix.arize.com/
- https://github.com/SynaLinks/synalinks
I saw an LLM introspection tool and website that look interesting here: https://www.neuronpedia.org/