Courses
Every course here is free. Some are bilingual readers of open-source material; each page names its source and licence.
A useful agent needs more than a model: tools, context, permissions and checks around it. This path starts with the simplest agent loop and ends with multi-runtime platform architecture.
Build a coding agent from the first loop up
Each session adds one piece to a coding agent's harness: tool use, permissions, hooks, skills, context compaction, memory, background tasks and agent teams. Every session comes with runnable Python code.
Make coding agents reliable and checkable
Lectures and projects on the environment, state, verification and control mechanisms that make AI coding agents dependable, such as why agents declare victory too early and how to stop it.
Take a coding agent from demo to production
A faithful bilingual reader for a build-a-coding-agent course: the outline, operating guides, architecture decision records and evaluation docs. Switch between Vietnamese and English on every page.
Compose agents from independent capabilities
A bilingual reader for the official Pydantic AI Harness documentation: the capability catalogue, every docs page and the guides for writing new capabilities.
From one agent to a multi-interface platform
A bilingual reader for Learn Hermes Agent in its original reading order: architecture overview, technical chapters across five stages, and data-structure references.
Building information modelling as a way of thinking, and hands-on Revit, from parametric concepts to a complete model and drawing set.
From BIM thinking to a complete building model
Five weeks of Revit. Each week has a concept lecture and step-by-step tutorials with chapter links to jump to the exact step, plus revision questions with answers.
