What's new

The biggest release yet. A full OpenAI side of the course to match the Anthropic side: a Codex & OpenAI tool section (models, the Responses API, the Codex CLI, and MCP), plus an OpenAI API in Practice cookbook (Batch API, caching, files & vision, tokens, hosted tools, Azure, the Agents SDK, parameters, and a Claude-to-OpenAI migration map). Every code-bearing lesson now shows Claude and OpenAI side by side on a toggle, so you can learn or migrate in either SDK. New too: a GitLab CI/CD with AI track showing how to automate GitLab merge-request review and pipelines with either model.

Every lesson and project page was rewritten so each concept is taught step by step — a stakes-first hook that shows why it matters, then a plain-English explanation with analogies, the underlying mechanism, and the common mistake to avoid. The goal: no page assumes you already know the thing it's teaching.

A self-hosted Python terminal landed on the site. Open any self-contained code block in a real in-browser shell, or use the full-page Playground (with numpy, pandas, scipy, and pydantic preloaded) — no install, no account, nothing leaves your machine.

The course launched at ai.studybydoing.in with 300+ pages across the full path from Python foundations to production LLM systems — free, no account. A client-side search box (sticky in the sidebar, typo-tolerant, on-site results only) makes any topic findable in a keystroke.