ai-engineering-from-scratch vs agents-towards-production
Two of the top projects in Learning and cookbooks, side by side: who leads on each score part and who each one suits, then setup, license, activity and what each review found.
ai-engineering-from-scratch
Free curriculum of 542 coding lessons on building AI systems from scratch
agents-towards-production
Jupyter tutorials for taking LLM agents from prototype to production
Verdict
ai-engineering-from-scratch ranks higher, 91 to 82. Of the 4 score parts we measure for both, ai-engineering-from-scratch leads on 2, agents-towards-production on none and 2 are even.
- Adoptionai-engineering-from-scratch leads by 18 points, 94 to 76: a wide gap.
- FreshnessEven at 100.
- Maintenanceai-engineering-from-scratch leads by 12 points, 95 to 83.
- Agent-readyEven at 45.
Pick ai-engineering-from-scratch if you are
Developers learning to build AI systems by writing the code themselves
- Lessons ship runnable code; some start dependency-free, e.g. a vectors.py linear algebra lesson
- Three staged projects with starters, reference solutions and local graders (project_test.py)
Pick agents-towards-production if you are
Engineers learning to deploy and operate LLM agents in production
- 28 tutorials covering memory, deployment, guardrails, GPU scaling, evaluation and UI
- Spans several stacks: LangGraph, Redis, Mem0, Arcade, Inngest, and Kotlin with JetBrains Koog
| What we compare | ai-engineering-from-scratch | agents-towards-production |
|---|---|---|
| Score parts, out of 100 | ||
| Adoption | 94, widely used | 76, popular |
| Freshness | 100, active | 100, active |
| Maintenance | 95, healthy | 83, healthy |
| Agent-ready | 45, minimal | 45, minimal |
| Facts from GitHub and the README | ||
| Stars | 79.4k | 21.9k |
| License | MIT (permissive) | custom license (read the license) |
| Last commit | Nov 2026 | Nov 2026 |
| Last release | Nov 2026 | Oct 2026 |
| Language | Python | Jupyter Notebook |
| Docker | No | No |
| GPU | Not stated | Optional |
| arm64 or Apple Silicon | Mentioned | Not stated |
ai-engineering-from-scratch
A 20-phase course covering math foundations, ML fundamentals, LLM engineering, agents and production infrastructure, with runnable lesson code in Python, TypeScript, Rust, Julia and Go. It includes three graded projects (retrieval evaluation, agent trace debugger, tool call firewall) with local graders, plus agent skills that turn a coding agent into a tutor. Lessons can be read on the website or run locally.
Who it is for: Developers learning to build AI systems by writing the code themselves
Strengths
- Lessons ship runnable code; some start dependency-free, e.g. a vectors.py linear algebra lesson
- Three staged projects with starters, reference solutions and local graders (project_test.py)
- Covers math through agents, MCP and production infrastructure across 20 phases
- Installable tutor skills via npx skills for Claude Code, Codex and other hosts
Weaknesses
- 542 lessons is a large scope; the README tells readers to pick one goal path
- Projects need different toolchains: Python 3.10+, Node.js 22.18+ and Rust
- Tutor skills require Node.js, npx and a skill-capable coding agent host
- Starter project code fails by design until you implement the stages
- Needs Python 3.10+, Node.js 22.18+ (agent trace project, tutor skills), Rust (tool call firewall project)
agents-towards-production
A collection of 28 tutorial notebooks on production concerns for GenAI agents: stateful workflows with LangGraph, memory (Redis, Mem0), web search APIs, FastAPI and Docker deployment, guardrails, GPU deployment, observability and evaluation. Many tutorials are contributed by vendors the README lists as sponsors. The README also promotes the author's paid course and newsletter.
Who it is for: Engineers learning to deploy and operate LLM agents in production
Strengths
- 28 tutorials covering memory, deployment, guardrails, GPU scaling, evaluation and UI
- Spans several stacks: LangGraph, Redis, Mem0, Arcade, Inngest, and Kotlin with JetBrains Koog
- Each topic has its own tutorials/ folder, so one can be read without the rest
Weaknesses
- Many tutorials are sponsor-contributed and built around the sponsor's hosted product
- Several tutorials need external services such as Tavily, Bright Data and RunPod
- README leads with course and newsletter promotion before the tutorial list
- License is not identified (NOASSERTION); no releases or versioning
- GPU optional
- Needs LangGraph, Redis, Mem0, Tavily, Bright Data, Contextual AI, Arcade, RunPod, Inngest
More in Learning and cookbooks
Building with these?
We help teams pick an open-source AI stack and run it in production. Client work never changes a rank.