Tech Stacks I Like: Free and Cheap Infrastructure for AI Experimentation

A living document of the free and cheap tools I use for AI work: GPU compute, hosting, model access, and agentic coding.

One model, multiple perspectives: Exploring Qwen3's task-aware embeddings

Exploring how Qwen3-Embedding reshapes its embedding space based on task instructions, with an interactive demo using the 20 Newsgroups dataset.

Why AI Isn't Sophisticated Copy-Paste

A response to claims that AI is just sophisticated copy-paste, examining the empirical evidence from IMO gold medals, novel algorithm discovery, and competitive programming.

Asynchronous calls to the OpenAI API

A tutorial on making non-blocking asynchronous calls to the OpenAI API in Python using asyncio, improving throughput when experimenting with LLMs.