AI Systems Engineering Course: RAG, Agents and LLMOps

AI Systems Engineering Beginner to Advanced is a free video course on how to design, build, secure, evaluate, deploy and scale modern AI systems. It is for developers and engineers who already know the basics of large language models and now want to build complete, production-grade AI applications: retrieval, agents, evaluation, security, observability and cost control.

The course is 21 videos and about 14.4 hours on YouTube: a short overview, then 20 numbered modules of roughly 35 to 47 minutes each. It moves in a clear line from foundations to full AI systems design, with a focus on real-world engineering and enterprise applications. See the full outline on the AI Systems Engineering course page.

Who this course is for

• Software developers moving into AI engineering who want a structured path rather than scattered tutorials.
• AI and ML engineers who can call a model but need the engineering around it: retrieval, orchestration, evaluation and reliability.
• Solution and platform architects designing AI applications for an organisation.
• Technical leads who review AI system designs and need a shared vocabulary for them.

The level is Intermediate. It helps to be comfortable with programming and with the idea of prompting an LLM before you start.

What you will learn

The 20 modules group into six themes.

1. Foundations, models and context (Modules 1 to 3). “AI Systems Engineering Foundations”, “Models, Inference and Routing” and “Context Engineering” cover the building blocks: how a model call fits into a system, how requests can be routed between models, and how to give the model the right context. These choices drive both quality and cost later on.

2. Data and retrieval (Modules 4 to 6). “Data and Ingestion Pipelines”, “Production RAG” and “Advanced Retrieval” cover getting your own data into an AI system and retrieving the right pieces at answer time. Retrieval quality is usually the biggest lever on answer quality in enterprise apps.

3. Agents (Modules 7 to 10). “Tool-Using Agents”, “Agent Orchestration”, “Agent Memory” and “MCP, A2A and Interoperability” move from a single model call to agents that use tools, coordinate with each other, remember state and connect to other systems through open protocols.

4. Quality, reliability and safety (Modules 11 to 14). “Evaluation Engineering”, “Reliability Engineering”, “AI Security Engineering” and “Guardrails and Responsible AI” cover how you know a system works, how you keep it working, and how you protect it and its users.

5. Architecture and operations (Modules 15 to 18). “AI Application Architecture”, “Observability and LLMOps”, “Cost and Performance Engineering” and “Deploying and Scaling AI Systems” are the production side: structure, monitoring, spend and scale.

6. Putting it together (Modules 19 and 20). “Enterprise Case Studies” and “AI Systems Design” apply everything to realistic scenarios, which is the best practice for design reviews and system design interviews.

How the course is organised

• 21 videos, about 14.4 hours in total.
• A 6-minute overview, then Modules 01 to 20 in order.
• Each module is a long-form lesson of roughly 35 to 47 minutes.
• The playlist runs in module order, so you can simply follow it from top to bottom.

Because each module is substantial, treat one module as one study session, with time afterwards to build a small piece yourself.

How to study it step by step

1. Watch the overview to see the full map of the 20 modules.
2. Modules 1 to 3: take notes on the core terms (inference, routing, context) and write down how each one applies to an app you know.
3. Modules 4 to 6: build a small retrieval pipeline over a few of your own documents, then improve it using the advanced retrieval ideas.
4. Modules 7 to 10: turn that pipeline into an agent that uses at least one tool, then add memory, and read up on how MCP and A2A connect agents to other systems.
5. Modules 11 to 14: write a small evaluation set for your agent, and list the security risks and guardrails it needs.
6. Modules 15 to 18: sketch the architecture for running it in production, including monitoring, cost limits and scaling.
7. Modules 19 and 20: use the case studies and the system design module to review your own design end to end.

At one module per day the course takes about three weeks; at two or three a week, plan for two months. The whole course is also a YouTube playlist.

For more hands-on agent work, continue with Building Production AI Agents with LangGraph or the MCP Server Engineering Masterclass.

Frequently asked questions

Is the AI Systems Engineering course free?

Yes. The complete course is free on YouTube. You can follow all 21 videos from the course page or the playlist.

How long does it take to finish?

The videos total about 14.4 hours. With time to build and practise between modules, most learners will need several weeks.

Is this course for beginners?

It is listed at Intermediate level. It starts with foundations, but you will get the most from it if you already program and have used an LLM API or chat tool.

Does it cover RAG and AI agents?

Yes. Modules 4 to 6 cover data pipelines, production RAG and advanced retrieval, and Modules 7 to 10 cover tool-using agents, orchestration, memory, and MCP, A2A and interoperability.