insideout Get Pro

Build it yourself

Real projects you build on your own machine, with your own tools.

AI engineering

How language models work, and how to build with them.

Classical Machine Learning A marketing model on real bank data Pick which bank customers to call. Neural Networks A digit reader from scratch Read handwritten digits with NumPy alone. Tokens and Embeddings A tokenizer and a bigram model on one book Split a book into tokens, then predict the next. Transformers A small GPT on Shakespeare Write a tiny GPT and train it on your laptop. Training and Tuning A small model fine-tuned with LoRA Teach a small model a new task with LoRA. Inference and Serving A serving benchmark on your laptop Time one model at three precisions. Prompting A support desk pipeline on a local model Turn support messages into clean JSON. Evaluation An evaluation report on two local models Show which of two models is better. Search and Retrieval Search over your own notes Search your notes by words and by meaning. Agents A database agent with traces and limits An agent that answers from a database. Agents A knowledge base your coding agent maintains Let your agent keep a wiki of what you read. Production AI A model service with a fallback Keep a model service up when things fail. Engineering Toolkit A small data API, from repo to container Ship a small API from repo to container.

Quant engineering

The maths and tools behind quant research and trading.

Each brief copies straight into Claude Code, Cursor or Codex. Your machine, free data, and your own model or key.