An AIOps Course: Build AI Agents That Find the Cause and Ask Before Fixing
Is there a hands-on AIOps course online?
Yes. Hands-On AIOps on AWS at systemdesign.academy is a self-paced course where you build AI agents for DevOps incident response: an agent that investigates with read-only tools, proposes one exact fix and waits for a person to approve it. It is in early access: 30 of 70 planned lessons are ready today, about 21 hours with 29 hands-on labs, and every later lesson is included.
Save your progress and get the AI Engineering cheat sheet PDF
This page explains what our AIOps course teaches, what you build, what is ready today and what is still being written, in 10 short answers. Every count on it is computed from the course's own lesson files.
10 questions · 4 sections · answers from the course
These answers come from our course Hands-On AIOps on AWS: Build AI Agents for DevOps: 30 lessons are ready to study today, about 21 hours of learning, with 29 hands-on labs that make real calls to an OpenAI model and more than 700 diagrams and real screenshots. New lessons are added as they are finished, and you get every one.

The short answer
You build an on-call AI agent that only reads, proposes one fix and waits for a person. Parts 1 and 2 are ready today as written lessons with real labs; the later parts are still being written.
1. What this AIOps course teaches
What AIOps means here, what you build, and who it is for.
What does AIOps mean in this course?
AIOps here means using AI agents for DevOps incident response. When something breaks, an agent reads the alerts, investigates the system with read-only tools, finds the likely cause and proposes one exact fix. A person approves or refuses it, and only then does plain code apply it. The course also measures how often the agent is right, because an agent you cannot measure is not one you can trust on call.
It is not a course about a dashboard product or about machine learning on metrics. You build the agent yourself, so you understand every step it takes.
What do you build in the course?
One whole system, a piece at a time. A demo shop that is broken on purpose (a web page, an API, a worker, Postgres, Redis and a NATS queue on a small Kubernetes in Docker). An on-call agent written in Python with LangChain and LangGraph. An approval flow where the model never applies a change itself. Evaluations that break the shop in known ways and count right answers, wrong actions and cost. Then the same system in production on AWS.
That is the full plan of 7 parts. Today Parts 1 and 2, which build agents from zero, are ready; the shop, the approval gate and the AWS production parts are still being written, as the next answers say.
Who is this AIOps course for?
Developers, DevOps engineers and SREs who already know Python, the command line and the basics of Git. You do not need to know AI or DevOps tools: Part 1 starts from what a language model really does, and an optional appendix teaches containers and Kubernetes for anyone who has not used them.
The free lesson "Before you start" lists what you should already know and the four tools to install, with a check that each one works.
2. What is ready today
The honest status of the course, counted from the published lessons.

How much of the AIOps course is ready?
30 lessons are ready to study today, of 70 planned (61 on the main path and 9 in the optional appendix). That is about 21 hours of learning, with 29 hands-on labs and 785 diagrams and real screenshots.
What you can study now: the opening lessons, Part 1 and Part 2. Part 1, Agents from zero, has all 13 lessons ready. Part 2, LangGraph from zero, has all 10 lessons ready. New lessons are added in course order as they are finished, and you get every one.
Full walkthrough: The full contents, with which lessons are ready
What do the ready lessons teach?
Part 1 builds an agent from zero: what a model does with tokens and the context window, messages and temperature, answers checked with Pydantic, what an agent is, one agent loop printed step by step, tools and why their output is untrusted, the loop written by hand in about 40 lines of Python, the same agent with LangChain's create_agent, an agent that reads real CloudWatch alarms on AWS, middleware for limits and human approval, decision models, and a cost log for every call.
Part 2 rebuilds it as a graph with LangGraph: state and reducers, routing and loops, hand-offs and parallel work, checkpoints that survive a real crash, pausing for a human with interrupt, timeouts and caching, memory, Deep Agents and tracing with LangSmith.
What is still being written?
Parts 3 to 7 and the appendix. In the plan, Part 3 puts the first agent on the real demo shop with an approval gate it cannot get around and an Ops Console. Part 4 moves the tools to MCP with a gateway. Part 5 grows one agent into an on-call team. Part 6 runs it in production on AWS, with Terraform, EKS and one IAM role per agent. Part 7 proves it works with a red team and a final evaluation.
We do not promise dates for these. They are published in order, and the contents page always shows what is ready.
3. Labs, hours and format
Where the hands-on work runs, and whether there is video.
What are the labs like, and do I need AWS?
29 ready lessons have a hands-on lab. The labs make real calls to an OpenAI model with your own key. 28 ready lessons come with a zip of their code. Nearly all the ready labs run on your own computer, with no servers: you pay only small model fees, and the lesson on the cost log shows what each run cost.
One ready lesson uses AWS: it sets up a budget alert and a read-only user, stages a real CloudWatch incident, and gives the agent two tools to find the cause. The production part on AWS is in the plan for Part 6, which starts with a budget and a one-command teardown.
Is this a video course?
Not today. It is a self-paced course of written lessons, with diagrams, real screenshots from real runs and code you run. 14 of the ready lessons are marked to get a video later, but no lesson plays a video on the site yet, so today you study from the text. Later parts come as text first.
Every lesson has the same order: the problem, why it matters, the idea with a diagram, the hands-on run, then a knowledge check.
4. Trying it and buying it
How to check the course yourself, and what one payment includes.
Can I try the AIOps course before I pay?
Yes. 3 lessons are free: "Introduction: what this course builds", "Before you start" and "What an AI agent is". "What an AI agent is" is a full Part 1 lesson with its lab, so you see the teaching and the code before you decide.
How does buying work?
You pay once and keep the course for life, with every lesson, code zip and video added to it later. There is no subscription. It sells at an early-access price while the later parts are written, and that price rises once every part is out. The price for your country is in the box on this page and on the course page. AIOps on AWS is its own course; it is not part of the System Design or AI Engineering course.
Full walkthrough: The course page, with the full contents and price
Where to go next
Everything on this page comes from Hands-On AIOps on AWS: Build AI Agents for DevOps. 30 lessons are ready to study today, about 21 hours of learning, with 29 hands-on labs that make real calls to an OpenAI model and more than 700 diagrams and real screenshots. New lessons are added as they are finished, and you get every one. Start with the free lessons before you decide.
These questions are a start. The interview asks more.
Get all 120 questions in the Agentic AI Engineering pack
Each question has why it is asked, a full answer, what the interviewer listens for, the follow-ups, the common mistakes and a diagram (153 diagrams in all). A PDF you keep, junior to staff level.
Pay by UPI, cards or netbanking through Razorpay. Pay by card or PayPal, in US dollars. Download at once from My packs.
