When NOT to Use
- Task is outside your authorization scope
- You need to implement controls (use implementing-* skills)
- Task is about analysis, not action (use analyzing-* skills)
- You don't have access to target systems
- Task requires compliance expertise (consult professionals)
- Task is about defense, not offense (use defensive skills)
Overview
AI Engineering from Scratch: a complete 473-lesson curriculum spanning 20 phases from math foundations to autonomous agent systems. Covers Python, TypeScript, Rust, and Julia. Built for agents and humans who need a structured path through modern AI engineering.
Source: rohitg00/ai-engineering-from-scratch
Install: npx skills add rohitg00/ai-engineering-from-scratch
Capabilities
- Navigate a structured 20-phase AI engineering curriculum
- Identify skill gaps by mapping current knowledge to phases
- Recommend targeted lessons based on learner level
- Run placement quizzes to find starting phase
- Run per-phase comprehension checks
When to Use
Trigger phrases:
"ai engineering curriculum"
"User wants to learn AI engineering from scratch"
"Need to assess someone's AI skill level"
"Looking for structured learning path in ML/DL/LLMs/Agents"
User wants to learn AI engineering from scratch
Need to assess someone's AI skill level
Looking for structured learning path in ML/DL/LLMs/Agents
Building AI training programs or onboarding materials
Searching for specific AI topic coverage
Phase Map
| Phase |
Topic |
Focus |
| 0 |
Setup |
Environment, tools, Python/TS/Rust/Julia config |
| 1 |
Math |
Linear algebra, calculus, probability, statistics |
| 2 |
ML Fundamentals |
Supervised/unsupervised, evaluation, pipelines |
| 3 |
Deep Learning |
Neural networks, backprop, CNNs, RNNs |
| 4 |
Computer Vision |
Image classification, detection, segmentation |
| 5 |
NLP |
Text processing, embeddings, sequence models |
| 6 |
Speech/Audio |
ASR, TTS, audio processing |
| 7 |
Transformers |
Attention, encoder-decoder, positional encoding |
| 8 |
Generative AI |
GANs, VAEs, diffusion models |
| 9 |
Reinforcement Learning |
Policy gradient, Q-learning, PPO |
| 10 |
LLMs from Scratch |
Tokenization, training, scaling laws |
| 11 |
LLM Engineering |
Fine-tuning, RAG, prompt engineering, evals |
| 12 |
Multimodal AI |
Vision-language models, cross-modal reasoning |
| 13 |
Tools/Protocols |
MCP, function calling, tool use patterns |
| 14 |
Agent Engineering |
ReAct, planning, memory, tool orchestration |
| 15 |
Autonomous Systems |
Self-improving agents, reflection, verification |
| 16 |
Multi-Agent/Swarms |
Agent coordination, delegation, consensus |
| 17 |
Infrastructure/Production |
Serving, monitoring, scaling, cost optimization |
| 18 |
Ethics/Safety/Alignment |
RLHF, red teaming, guardrails, interpretability |
| 19 |
Capstone |
End-to-end project combining all phases |
Built-in Agents
- Primary agent handles core task execution
- Validator agent checks output quality
- Reporter agent formats and delivers results
- Each agent operates with clear input/output contracts
/find-your-level
Placement quiz that assesses current knowledge across phases and recommends a starting point.
/check-understanding
Per-phase quiz that tests comprehension after completing each phase.
Usage
User: "Where should I start learning AI?"
Agent: Runs /find-your-level placement quiz, maps results to phase map, recommends starting phase
User: "I know Python and basic ML, what next?"
Agent: Maps to Phase 3-4, recommends deep learning and computer vision modules
User: "Quiz me on transformers"
Agent: Runs /check-understanding for Phase 7
How to Use
- Invoke the skill when relevant domain keywords appear in the request
- Provide required inputs as specified in the skill definition
- Review the output for correctness before delivering to the user
- Combine with related skills for complex multi-step workflows
Verification
After completing this skill, confirm:
Process
- Analyze the task requirements
- Apply domain expertise
- Verify output quality
Anti-Rationalization Table
| Rationalization |
Reality |
| "I will add monitoring later" |
Without monitoring, you cannot detect failures. Add it from day one. |
| "One model is enough" |
Different tasks need different models. Route intelligently. |
| "Premature optimization" |
Infrastructure decisions are hard to change later. Design for scale early. |
1---2name: ai-engineering-curriculum3description: Use when structured AI engineering curriculum — 382 skills + 99 prompts across 20 phases covering ML, deep learning, LLMs, agents, and production systems. Use when learning AI, building AI skills,.4license: Apache-2.05---678## When NOT to Use910- Task is outside your authorization scope11- You need to implement controls (use implementing-* skills)12- Task is about analysis, not action (use analyzing-* skills)13- You don't have access to target systems14- Task requires compliance expertise (consult professionals)15- Task is about defense, not offense (use defensive skills)161718## Overview1920AI Engineering from Scratch: a complete 473-lesson curriculum spanning 20 phases from math foundations to autonomous agent systems. Covers Python, TypeScript, Rust, and Julia. Built for agents and humans who need a structured path through modern AI engineering.2122Source: [rohitg00/ai-engineering-from-scratch](https://github.com/rohitg00/ai-engineering-from-scratch)2324Install: `npx skills add rohitg00/ai-engineering-from-scratch`2526## Capabilities2728- Navigate a structured 20-phase AI engineering curriculum29- Identify skill gaps by mapping current knowledge to phases30- Recommend targeted lessons based on learner level31- Run placement quizzes to find starting phase32- Run per-phase comprehension checks3334## When to Use3536**Trigger phrases:**37- "ai engineering curriculum"38- "User wants to learn AI engineering from scratch"39- "Need to assess someone's AI skill level"40- "Looking for structured learning path in ML/DL/LLMs/Agents"414243- User wants to learn AI engineering from scratch44- Need to assess someone's AI skill level45- Looking for structured learning path in ML/DL/LLMs/Agents46- Building AI training programs or onboarding materials47- Searching for specific AI topic coverage4849## Phase Map5051| Phase | Topic | Focus |52|-------|-------|-------|53| 0 | Setup | Environment, tools, Python/TS/Rust/Julia config |54| 1 | Math | Linear algebra, calculus, probability, statistics |55| 2 | ML Fundamentals | Supervised/unsupervised, evaluation, pipelines |56| 3 | Deep Learning | Neural networks, backprop, CNNs, RNNs |57| 4 | Computer Vision | Image classification, detection, segmentation |58| 5 | NLP | Text processing, embeddings, sequence models |59| 6 | Speech/Audio | ASR, TTS, audio processing |60| 7 | Transformers | Attention, encoder-decoder, positional encoding |61| 8 | Generative AI | GANs, VAEs, diffusion models |62| 9 | Reinforcement Learning | Policy gradient, Q-learning, PPO |63| 10 | LLMs from Scratch | Tokenization, training, scaling laws |64| 11 | LLM Engineering | Fine-tuning, RAG, prompt engineering, evals |65| 12 | Multimodal AI | Vision-language models, cross-modal reasoning |66| 13 | Tools/Protocols | MCP, function calling, tool use patterns |67| 14 | Agent Engineering | ReAct, planning, memory, tool orchestration |68| 15 | Autonomous Systems | Self-improving agents, reflection, verification |69| 16 | Multi-Agent/Swarms | Agent coordination, delegation, consensus |70| 17 | Infrastructure/Production | Serving, monitoring, scaling, cost optimization |71| 18 | Ethics/Safety/Alignment | RLHF, red teaming, guardrails, interpretability |72| 19 | Capstone | End-to-end project combining all phases |7374## Built-in Agents75- Primary agent handles core task execution76- Validator agent checks output quality77- Reporter agent formats and delivers results78- Each agent operates with clear input/output contracts798081### /find-your-level82Placement quiz that assesses current knowledge across phases and recommends a starting point.8384### /check-understanding85Per-phase quiz that tests comprehension after completing each phase.8687## Usage8889```90User: "Where should I start learning AI?"91Agent: Runs /find-your-level placement quiz, maps results to phase map, recommends starting phase9293User: "I know Python and basic ML, what next?"94Agent: Maps to Phase 3-4, recommends deep learning and computer vision modules9596User: "Quiz me on transformers"97Agent: Runs /check-understanding for Phase 798```99100## How to Use1011021. Invoke the skill when relevant domain keywords appear in the request1032. Provide required inputs as specified in the skill definition1043. Review the output for correctness before delivering to the user1054. Combine with related skills for complex multi-step workflows106107## Verification108109After completing this skill, confirm:110111- [ ] Output meets the defined quality and completeness requirements112- [ ] All prerequisites are verified and documented113- [ ] Error handling covers edge cases114- [ ] Results are accurate and actionable115116## Process1171181. Analyze the task requirements1192. Apply domain expertise1203. Verify output quality121122## Anti-Rationalization Table123124| Rationalization | Reality |125|---|---|126| "I will add monitoring later" | Without monitoring, you cannot detect failures. Add it from day one. |127| "One model is enough" | Different tasks need different models. Route intelligently. |128| "Premature optimization" | Infrastructure decisions are hard to change later. Design for scale early. |
oyi77