百科.dev
全部条目在线工具AI 编程趋势榜开源项目技术资讯提交条目
登录
< 返回工具列表
A

ai-builders-curriculum

> 编程语言
开源

开放、供应商中立的课程和黑客马拉松启动包,用于学习构建端到端的全栈 AI 应用。采用 MIT 许可证。AI Builde 项目

390 stars0 点赞1 次浏览
访问官网GitHub

工具介绍

开放、供应商中立的课程和黑客马拉松启动包,用于学习构建端到端的全栈 AI 应用。采用 MIT 许可证。AI Builde 项目

AI Builders Curriculum

Learn to build and ship a real, full-stack AI application. Six short modules and three runnable starter kits take you from an empty folder to a deployed app with a database, real user accounts, an API, and an AI feature you can see and govern — free, open source, and taught with a vendor-neutral stack so the skills transfer no matter which tools you end up using.

Build this in a weekend

Three small, readable starter kits, each a complete app, not a toy demo. Clone one and it's running locally in about a minute, with no API key required to try it.

️ ai-app-starter — a full-stack AI app

A notes app with a database, an HTTP/JSON API, and an LLM-powered Summarize button.

bash
git clone https://github.com/ai-builders-foundation/ai-builders-curriculum.git
cd ai-builders-curriculum/starter-kits/ai-app-starter
npm install && npm start

Open http://localhost:3000, write a note, click Summarize, and watch the backend call a language model. It ships in offline demo mode by default, so this works with zero configuration — point it at any OpenAI-compatible endpoint later by editing three environment variables. Nothing else in the code changes.

rag-starter — a "chat with your docs" app

Paste in text, ask a question, and get an answer grounded in only what you pasted — with the exact source chunk cited.

bash
cd ../rag-starter        # from the same clone, or start fresh
npm install && npm start

Open http://localhost:3000 and ask "What are the five steps of RAG?" — the answer comes back grounded in a sample document that's seeded automatically, with the retrieved chunk shown as a source. Again, no API key needed to see it work end to end.

glassbox: the same RAG app, with its hidden decisions on screen

rag-starter above quietly hardcodes six real decisions: how many sources it retrieves, how creative it's allowed to be, how long an answer can run, what tone it writes in, how it cites sources, and what it does when it doesn't know the answer. glassbox is that exact app with all six lifted out of the code and turned into named, explained, adjustable controls. Every answer reports the full set of decisions that produced it, defaults included, not just what you changed.

bash
cd ../glassbox         # from the same clone, or start fresh
npm install && npm run migrate && npm start

Open http://localhost:3000, ask a question, then flip whenUnsure to answer-anyway and ask something the seeded documents can't answer. Watching a real model go from "I don't know" to confidently making something up, with one setting changed, is the fastest way to feel why grounding matters. No API key needed for the core experience. Read alongside 06: Transparent AI, and see BUILD-ALONG.md for a ready-to-run 90-minute session plan.

Every kit here is small enough to read in one sitting: plain Node.js, SQLite, no framework, no build step, and is the concrete, runnable version of the curriculum below.

The learning path

** Data → Auth → Functions/APIs → AI & Agents → Deploy → Transparent AI**

# Module You'll learn to…
01 Data & Storage Model data, choose a database, write schema & migrations, query safely
02 Auth & Users Hash passwords, run sessions, protect routes, understand OAuth & roles
03 Functions & APIs Build HTTP endpoints, validate input, handle errors, design a clean API
04 AI & Agents Call an LLM behind an env var, get structured output, use tools, run an agent loop
05 Deploy Package the app, manage secrets, ship to a host, add health checks & logging
06 Transparent AI Find an AI app's hidden decisions, name and bound them, judge which ones to expose

Each module: learning objectives → explanation with worked examples → hands-on exercise → "Check yourself." Budget one module per evening, or two per weekend.

Looking for more free material on any of this? See RESOURCES.md — a curated list of the best free resources for each topic above.

Who it's for

  • Self-learners who can write a little code and want to build a real AI app end to end.
  • Educators & bootcamps who want an open, adaptable syllabus to teach from directly — it's MIT, fork it.
  • Hackathon organizers who want a shared, runnable baseline so participants spend the weekend building, not configuring.

Prerequisites: basic programming (variables, functions, loops), a terminal, and Node.js 18+. No prior AI, backend, or deployment experience needed — that's what this teaches.

Get involved

⭐ Star this repo if it's useful — it helps other builders find it.

PRs are genuinely welcome: a fixed typo, a clearer explanation, a new exercise, a translation, or a whole new module or starter kit. Start with CONTRIBUTING.md.

Repository layout

…

Maintained by the AI Builders Foundation, a 501(c)(3) public charity (EIN 42-1986706), whose mission is to educate and support AI builders through free courses, community events, and open source. This repository is those three pillars as code — free forever, with no paid tier and no lock-in to any platform, including our own.

MIT License · Code of Conduct · Security Policy · [email protected]

Issues· 0 开放

查看全部 Issues在 GitHub 打开

暂无开放 Issues,或尚未同步最近议题。

> 标签

TypeScript

暂无评论,来聊聊你的看法吧

> 工具信息

发布日期2026年8月1日
最后更新2026年9月17日
分类编程语言
定价开源

> 相关工具

T
TypeScript
JavaScript 的超集,为前端与全栈提供静态类型
P
Python
通用编程语言,广泛用于 Web、数据与 AI
G
Go
Google 推出的简洁高效系统语言