A curriculum and field guide for building, evaluating and operating AI systems.
Open the interactive site index — every section, course lecture, study aid, and all 251 files of this repository on one searchable page, styled as a companion to the AI-Safety site. Works on phone and desktop.
What this repository is: a free, self-study library for becoming an AI engineer. It has three layers — a technical curriculum (sections 101–106, numbered in learning order), roadmaps (guided, day-by-day study plans through that same material), and interview & career material — all written from first principles so you can start from zero.
Prerequisites: you can already program (Python is the most useful language here). No machine-learning background is required — section 101 teaches every concept from scratch.
| Your situation | Start here | What it is |
|---|---|---|
| "I'm completely new to LLMs and AI engineering" | 101 — LLM Literacy | The on-ramp: how LLMs work, prompting, context engineering — from zero |
| "I want a day-by-day plan to become job-ready" | Accelerated Roadmap (100 days) | 6 modules · 100 days · the core ~80% employers expect |
| "I want full depth, at my own pace" | Full Roadmap | 16 levels · 286 sessions · 100% coverage from software fundamentals to interview prep |
| "I need to learn one specific topic" | Curriculum sections 101–106 | Reference knowledge base — open just the section you need |
| "I'm switching from another engineering role" | Learning Paths | Transition guides from data / backend / frontend / ML engineering, with durations |
| "I have an interview coming up" | Interview Preparation | The process, 100 answered questions, system design, negotiation |
The six numbered sections build on each other — each assumes roughly what the previous ones taught, so read them in order the first time and use them as reference afterwards:
| # | Section | You will learn… |
|---|---|---|
| 101 | LLM Literacy | how LLMs actually work — and how to steer them with prompts and context |
| 102 | AI Application Development | building applications: orchestration, vector databases, RAG, memory, fine-tuning |
| 103 | Agentic AI | autonomy: planning loops, tool protocols (MCP), agent frameworks, multi-agent systems |
| 104 | LLMOps | production: deployment, observability, safety guardrails, governance |
| 105 | AI Evaluation | evidence: proving your system works — includes a full 19-lecture course |
| 106 | AI Security & Safety | defense: threat modeling, prompt-injection defense, red teaming, alignment, responsible AI |
Roadmaps vs. curriculum: the roadmaps are the guided tours — they tell you what to study and in what order. The curriculum sections are the reference library beneath them — deep notes on each topic. Follow a roadmap as your schedule, and open the matching curriculum section whenever you want the full explanation.
Two complete roadmaps for becoming an AI engineer, plus a full interview-preparation track — structure adapted verbatim from the CampusX AI Engineering Roadmap 2026, with every learning resource replaced by free worldwide alternatives:
- Roadmap hub — which one to follow — comparison, and how important each topic is for interviews vs the real-world job
- Full Roadmap — 16 levels · 7 phases · 286 sessions · 14 projects; 100% coverage from software fundamentals to interview prep
- Accelerated Roadmap — 100 Days of AI Engineering — 6 modules · 100 days; the core ~80% you need to land your first AI engineering job
- The 100 GenAI / AI Engineer Interview Questions — copied exactly, with strong-answer outlines and a practice loop
- 12 portfolio projects — full specifications with "done when" criteria and a target-role selection table
- 101 — LLM Literacy — first-principles guide to transformer mechanics, inference sampling, prompt & context engineering
- 102 — AI Application Development — orchestration, vector databases, complete RAG curriculum, memory systems, and fine-tuning
- 103 — Agentic AI — autonomous planning loops, Model Context Protocol (MCP), agentic frameworks, and multi-agent orchestration
- 104 — LLMOps — deployment architectures, observability, safety guardrails, governance, and enterprise scaling
- 105 — AI Evaluation — evaluation fundamentals, LLM-as-a-Judge, agent benchmarks and the integrity crisis, RAG evaluation practice, eval engineering in CI/CD, dataset engineering, and online experimentation — plus a complete 19-lecture LLM Evaluations course with full English notes and transcripts, a glossary, a self-assessment quiz, and a cheat sheet
- 106 — AI Security & Safety — threat modeling (OWASP LLM + Agentic Top 10, MITRE ATLAS), prompt injection defense, red teaming, agent & MCP security, AI alignment, responsible AI governance, the security & safety dataset catalog, and worldwide incident databases
- My vision of the role — how I see AI engineering, comparison with DS/ML/DE roles, CRISP-DM for AI
- Skills analysis — reproduced annotation counts, limitations and a competency rubric
- Responsibilities — engineering ownership and reviewable deliverables
- Use cases — application patterns, data requirements and measurable outcomes
- Reality vs. job postings — what candidates experience vs. what's advertised
- Worldwide job market analysis — global geographical demand, compensation profiles, and company archetypes
- Interview process — common patterns, step counts, time estimates, AI use in hiring, key takeaways
- Interview questions — practice topics and individually attributed legacy reports — every question now carries an expert answer (
**A:**)- Theory — LLMs, RAG, agents, context engineering, evaluation, safety, fine-tuning, MCP — answered
- Coding — coding round formats, DSA problems, from-scratch ML implementations — answered
- Project deep dive — presentation rounds, follow-up probes, what interviewers evaluate — answered
- AI system design — system design for AI applications — answered
- Behavioral — values, leadership, problem-solving — answered
- Home assignments — assignment preparation and public examples — answered
- Skills that get you hired — baseline expectations, differentiators, and portfolio strategy
- After the interview — handling offers, rejections, and salary negotiation
- Interview trends — AI cheating, AI-proctored rounds, context engineering interviews
- Global compensation & negotiation guide — compare dated local evidence and written offer terms
- Resume & portfolio playbook — ATS optimization, high-impact bullet formulas, and 60-second GitHub repo audit checklist
- Production system design case studies — illustrative design exercises and validation requirements
- General learning path — what to learn and in what order
- From foundations to advanced engineering — competency stages, real-data projects, measurable exit criteria and expert review questions
- From Data Engineer — smoothest transition, 3-4 months
- From Data Scientist — evaluation is your superpower, add engineering
- From ML Engineer — easiest transition, replace model call with API call
- From Backend Engineer — 2-3 months, add AI on top of engineering
- From Frontend Engineer — backend first, then AI, unique full-stack advantage
- Project ideas and built projects — project specifications with evaluation guidance, plus the five built project repositories (101–105)
Curated collection of resources compiled while researching content for this field guide:
- Practitioner interview stories
- AI system design guides
- Company engineering blogs
- Books and courses
- Case study collections
See awesome.md for the list.
Supplemental notes on AI engineering careers. Recording, speaker and survey claims in these notes have not been independently verified; see their evidence status.
- A Day of an AI Engineer — the practical reality of the role
- Defining the AI Engineer Role — role scope and engineering responsibilities
- The Interview Process — real hiring trends, technical questions, and live coding challenges
- Take-Home Assignments — analyzing real assignments and building production-ready solutions
How the claims in this guide are sourced, checked, and recorded:
Evidence review: October 9, 2026. The latest job-data snapshot inspected is September 23, 2026. Review dates and source publication/collection dates are separate. Some long-form legacy claims still need source-level verification; see the audit.
Section review: October 27, 2026. The AI Evaluation and AI Security & Safety sections were added with all references verified; the security section was extended on this date with a dataset catalog and incident databases, all figures verified against live sources.
- Evidence and real datasets — a reproduced job-posting analysis, dataset limitations, dated benchmark references and reproduction commands.
- Evidence policy — distinguish measurements, source-reported claims, design targets and teaching illustrations.
- Repository audit — what was checked, corrected and left unresolved.
The reproduced source contains 8,051 stored job observations across nine scrape folders, including 1,086 records in the September 23, 2026 folder. These are counts from a selected job-board dataset, not worldwide demand estimates. Structured skills and role labels are LLM-derived annotations; the analysis and input fingerprint make the counting reproducible without certifying every annotation.
Advanced practice requires a baseline, versioned data, held-out evaluation, failure analysis, authorization checks, measured operating costs and a recovery plan. The curriculum retains its introductory explanations; the advanced path connects them to reviewable engineering artifacts.
This guide builds on the foundational work by Alexey Grigorev in the original AI Engineering Field Guide. The October 9 review links to a pinned upstream dataset and adds a reproducible counting method. Upstream annotations and third-party data retain their own limitations and rights; this repository does not claim a newly collected worldwide dataset.
Building • Learning • Researching • Sharing
⭐ If this repository helped you, consider giving it a star! ⭐
Engineering principle: Robust AI engineering requires deterministic software engineering surrounding probabilistic model outputs. Focus on evaluation suites, reproducible benchmarking, and production monitoring to ensure reliable business outcomes.
Open the interactive site index — every note, course, and guide in this repository on one page.