AI Engineering Roadmap by CampusX

AI Engineering Roadmap

Sixteen levels from writing software to shipping AI systems.

Learn AI engineering step by step. Each level builds on the ones before it, and most end with a project you build.

See the Roadmap How to use it
The path at a glance 16 levels · 286 sessions
  1. Foundations 0 1 2 3
  2. Retrieval 4
  3. Agentic 5 6
  4. Evaluation & Trust 7 8
  5. Production 9 10 11
  6. Frontier & Proof 12 13 14
  7. Interview 15

Go through the phases in order

How to use it

Three simple rules.

  1. 01

    Start where you are

    Each level lists what you need to know first. If you already know it, skip ahead.

  2. 02

    Do one level at a time

    Each level has a few modules, and most end with a project. Go in order, because each level builds on the last.

  3. 03

    Finish by building

    A level is done when you've built its project and can do everything on its outcome list. Watching the videos is not enough.

The roadmap

All 16 levels, in order.

16 levels in 7 phases. Each color is a phase. Click any level to see its modules and project.

The whole path 16 levels · 286 sessions
  • Foundations 0–3
  • Retrieval 4
  • Agentic 5–6
  • Evaluation & Trust 7–8
  • Production 9–11
  • Frontier & Proof 12–14
  • Interview 15
Phase 1

Foundations

Levels 0–3

software, LLM 101, orchestration, prompting

67 sessions

  1. 0 Level 0 Software Engineering Fundamentals Become comfortable building, debugging, and deploying real software before adding LLMs. This is the engineering foundation for everything that follows.
    22 sessions · 8 modules
  2. 1 Level 1 LLM 101 Build a strong mental model of how LLMs work, why they behave the way they do, and how to choose and use the right model for a task.
    15 sessions · 12 modules
  3. 2 Level 2 Working with LLM APIs (Orchestration) Turn individual LLM calls into reliable software workflows — messages, streaming, structured outputs, tools, multi-step orchestration, concurrency, failure handling, and observability. RAG and agents come later.
    19 sessions · 12 modules
  4. 3 Level 3 Prompt Engineering Design instructions that make LLM behavior more reliable and predictable. Context engineering comes later.
    11 sessions · 9 modules
Phase 2

Retrieval

Level 4

embeddings, vector search, RAG

27 sessions

  1. 4 Level 4 RAG Build the retrieval layer that RAG depends on — ingestion, chunking, embeddings, vector search, and hybrid retrieval — then ground LLMs in external knowledge, improve retrieval quality, and learn the major RAG architectures used beyond a naive retrieve-and-generate pipeline.
    27 sessions · 15 modules
Phase 3

Agentic

Levels 5–6

agents, context engineering

35 sessions

  1. 5 Level 5 AI Agents Build systems where models can decide what to do next, use tools, maintain state, recover from failures, and collaborate with other agents. Planning · tools · MCP · memory · multi-agent · LangGraph.
    22 sessions · 11 modules
  2. 6 Level 6 Context Engineering Manage the model’s limited context window deliberately — decide what enters, what stays, what gets compressed, and what gets moved outside — and build the agent harness that applies these decisions across long-running work.
    13 sessions · 9 modules
Phase 4

Evaluation & Trust

Levels 7–8

evals, security

32 sessions

  1. 7 Level 7 LLM Evaluations Design, execute, and automate comprehensive evaluations for Large Language Models and AI applications to ensure quality, safety, and reliability in production environments.
    17 sessions · 9 modules
  2. 8 Level 8 AI Security Secure LLM, RAG, and agent systems against model-specific attacks, unsafe actions, data leakage, and regulatory risk.
    15 sessions · 9 modules
Phase 5

Production

Levels 9–11

system design, LLMOps, managed platforms

57 sessions

  1. 9 Level 9 AI System Design Design a complete production AI application and defend every architectural choice across quality, latency, cost, reliability, and user experience.
    16 sessions · 10 modules
  2. 10 Level 10 LLMOps: From Development to Production Take the RAG application and the agent you have already built into production — versioned, evaluated on every change, guarded, containerized, deployed on AWS, and monitored — and keep them improving once real users arrive.
    23 sessions · 11 modules
  3. 11 Level 11 Managed AI Platforms Ship an AI application on the platforms teams actually buy — a cloud provider's managed platform, an open-weight hosting provider, or a router spanning several — and know which parts to hand over and which to keep.
    18 sessions · 10 modules
Phase 6

Frontier & Proof

Levels 12–14

multimodal, fine-tuning, projects

52 sessions

  1. 12 Level 12 Multimodal 101: Beyond Text Build a strong mental model of how models handle images, audio, and video, why they fail differently from text models, and how to choose and use the right model when a task stops being text-only.
    20 sessions · 13 modules
  2. 13 Level 13 Fine-Tuning Fine-tune a model only when prompting, retrieval, and orchestration have measurably run out, and prove with evaluation numbers that the tune was worth its data, training, and serving cost.
    16 sessions · 10 modules
  3. 14 Level 14 Projects Proof, not concepts. Portfolio builds that compose earlier levels. Ship 3–5 public.
    16 sessions · 7 modules
Phase 7

Interview

Level 15

positioning, scenarios, mock loops

16 sessions

  1. 15 Level 15 GenAI / AI Engineer Interview Prep Convert everything learned across the roadmap into interview performance. This level does not re-teach concepts; it focuses on scenario-based questions, system-design discussions, debugging cases, trade-offs, and defending real engineering decisions under interview conditions.
    16 sessions · 10 modules

What you'll build

14 projects you'll build.

Most levels end with a project. Together, they make your portfolio.

CampusX One

The roadmap is free. Learn it on CampusX One.

Most levels have matching courses on CampusX One. Use the roadmap to decide what to learn next, and CampusX One to learn it.

Explore CampusX One Start at Level 0
  • Foundations L0–3
  • Retrieval L4
  • Agentic L5–6
  • Evaluation & Trust L7–8
  • Production L9–11
  • Frontier & Proof L12–14
  • Interview L15