Agentic AI for DevOps

Build AI agents that automate real DevOps infrastructure tasks — from kubectl error explanations to a self-healing Kubernetes system.

A 2-day hands-on course by Shubham Londhe / TrainWithShubham

---

The Journey

You start with zero LLM experience and end up building KubeHealer — an AI agent that detects broken Kubernetes pods, diagnoses the root cause, and fixes them automatically.

Day 1 — Foundations

Module	What You'll Build
Module 0 — Know Before You Go	Set up your environment and verify everything works
Module 1 — Docker Error Explainer	Paste a Docker error, get a human-readable fix (your first LLM tool)
Module 2 — Docker Troubleshooter Agent	An AI agent that inspects containers and diagnoses crashes on its own
Module 3 — Multi-Tool DevOps Agent + MCP	Combine Docker + K8s tools, intro to LangChain and MCP

Day 2 — Production-Grade Agents

Module	What You'll Build
Module 4 — AIOps Demystified	The AIOps landscape, guardrails, and why durability matters
Module 5 — KubeHealer	Self-healing K8s agent with Temporal and Claude (separate repo)
Module 6 — Build Your Own AIOps Agent	CI/CD Failure Analyzer — diagnose GitHub Actions failures with an AI agent

---

Prerequisites

Tool	Why
Docker	We build and troubleshoot containers
kubectl	We talk to Kubernetes clusters
Kind	Local K8s clusters for testing
Python 3.10+	All code is Python
Ollama	Run LLMs locally, no API keys needed
Temporal CLI	Durable workflow execution (Module 5)
gh CLI	GitHub CLI for CI/CD analysis (Module 6)
Anthropic API key	Claude AI for KubeHealer (Module 5)

Already have Docker, kubectl, and Kind? Head to Module 0 to set up the rest.

---

Quick Start

git clone https://github.com/trainwithshubham/agentic-ai-for-devops.git
cd agentic-ai-for-devops

python3 -m venv .venv
source .venv/bin/activate
pip install -r requirements.txt

python3 module-0/verify_setup.py

See Module 0 for the full step-by-step setup.

---

Tech Stack

Category	Tools
Language	Python 3.10+
LLM (Day 1)	Ollama + Gemma 4 (local, free)
LLM (Day 2)	Anthropic Claude (KubeHealer)
Containers	Docker
Orchestration	Kubernetes (Kind)
Agent Framework	LangChain
Durable Execution	Temporal
MCP	FastMCP

---

Built by Shubham Londhe for the TrainWithShubham community