An open guide for university students and junior researchers worldwide
English · 中文 · Project repository
Build AI foundations, complete reproducible projects, and learn to turn a question into research that others can examine. The guide connects programming, mathematics, machine learning, research areas, literature review, experimental design, evaluation, engineering, and academic communication. English and Chinese editions share the core learning routes and resources.
| Goal | Guide |
|---|---|
| Plan systematic study | Learning path · Foundations |
| Choose structured courses and exercises | Courses · Practice |
| Complete a first project | Starter projects |
| Choose a research question | Research map · Specialized topics |
| Read, design, and evaluate research | Reading papers · Experimental design · Datasets |
| Produce reliable research artifacts | Research engineering · Responsible AI |
| Communicate and collaborate | Writing · Academic contact |
The 60-question research handbook answers common questions across these stages. The resource index provides the full directory. Attributed personal experiences remain supplementary reading.
Use Node.js 22 or newer and Python 3 for the local server.
npm ci
npm run build
npm run devOpen the English presentation homepage or the Chinese homepage.
| Page | Generated file |
|---|---|
| Repository root for GitHub Pages | index.html |
| English presentation homepage, the default local entry | public/blog/guide/index.html |
| Chinese presentation homepage | public/blog/guide/zh.html |
| Site root, also showing the presentation homepage | public/index.html |
| English guide overview | public/blog/guide/en/index.html |
| Chinese guide overview | public/blog/guide/zh/index.html |
This revision is a local preview. Existing public deployments may show earlier content; publishing is a separate step. External links have not all been individually rechecked. Resource pages explain their verification scope and dates.
npm run build # Main site, presentation homepages, and repository-root entry
npm run build:pages # GitHub Pages package in dist/github-pages
npm run build:uestc # Secondary edition
npm run build:uestc:standalone # Standalone package of the secondary edition
npm test # Build and behavior regression checksBoth editions use the core Markdown in content/guide/en and content/guide/zh. The main edition contains 45 pages per language, including three community perspective pages. The annual review (timeline) is no longer published; its Markdown sources are retained as an archive. The secondary manifest selects 42 pages per language and excludes those community essays; its guide is generated at public/blog/guide/guide-uestc/en/index.html and the corresponding zh/index.html. Historical Markdown overrides in content/guide/uestc remain in the repository and are no longer used as separate core articles.
npm run build generates a complete English homepage as the repository-root index.html, together with .nojekyll. Its links and assets use relative paths under public/blog/guide/, so it also works when the site is hosted under a repository name.
Choose a publishing method in Settings → Pages → Build and deployment:
- Deploy from a branch: run
npm run build, commit the rootindex.html,.nojekyll, and the completepublic/blog/guide/directory, then select the publishing branch and /(root). Rebuild and commit the generated output whenever the source content changes. - GitHub Actions: select GitHub Actions as the source. The included Pages workflow runs
npm run build:pagesand publishesdist/github-pages/. This package contains its own rootindex.html,.nojekyll, andpublic/blog/guide/directory for the main edition.
Keep the full directory structure when uploading or copying a build; index.html alone is not a complete website. The local changes here have not been published to a remote repository. See GitHub's publishing-source instructions for the repository settings.
Maintained by OpenEnvision, this project builds on Takamatsu-Hikaru/AI-Research-Guide, earlier student-contributed AI learning materials. Original and adapted articles retain author credit and links to their sources. The Lumina Embodied AI Guide informed embodied-AI resources; irene provided the source experience for the paper-reading workflow.
See About this guide for editorial standards and acknowledgments, and CONTRIBUTING.md for editing guidance. Propose corrections with the affected page, supporting original source, and corresponding language update. Field material is in content/guide/fieldnotes; figure attribution is recorded in paper-figures.json.