AI for IT Professionals
— Complete Notes

# Install on Linux
curl -fsSL https://ollama.com/install.sh | sh

# Install on macOS
# Download from: https://ollama.com/download/mac

# Pull and run a model (downloads automatically)
ollama run llama3.2         # Meta's Llama 3.2 — great general model
ollama run mistral          # Mistral 7B — fast, efficient
ollama run phi3             # Microsoft Phi-3 — small but smart
ollama run codellama        # Optimised for code generation

# List downloaded models
ollama list

# Run in API mode (works like OpenAI API)
ollama serve                # starts on localhost:11434

# Call it with curl
curl http://localhost:11434/api/generate -d '{
  "model": "llama3.2",
  "prompt": "Explain what a Docker container is in 2 sentences",
  "stream": false
}'

# Install Open WebUI — browser interface for Ollama
docker run -d -p 3000:8080 \
  -v open-webui:/app/backend/data \
  --add-host=host.docker.internal:host-gateway \
  ghcr.io/open-webui/open-webui:main
# Then open: http://localhost:3000

# docker-compose.yml — Ollama + Open WebUI
services:
  ollama:
    image: ollama/ollama
    ports:
      - "11434:11434"
    volumes:
      - ollama:/root/.ollama
    # Add GPU support if you have NVIDIA:
    # deploy:
    #   resources:
    #     reservations:
    #       devices:
    #         - driver: nvidia
    #           count: all

  open-webui:
    image: ghcr.io/open-webui/open-webui:main
    ports:
      - "3000:8080"
    environment:
      - OLLAMA_BASE_URL=http://ollama:11434
    volumes:
      - open-webui:/app/backend/data
    depends_on:
      - ollama

volumes:
  ollama:
  open-webui:

# Run it:
docker compose up -d
# Open: http://localhost:3000
# Pull a model: docker exec -it ollama ollama pull llama3.2

── Pattern 1: Role + Task + Context ──
"You are a senior DevOps engineer with 10 years of Linux experience.
I have a bash script that backs up /var/www to S3 using aws-cli.
It works manually but fails silently when run via cron.
Diagnose the most likely causes and give me fixes for each."

── Pattern 2: Step by Step ──
"Explain how Docker networking works.
Do it step by step. Start with the simplest concept and build up.
Use an analogy before any technical terms."

── Pattern 3: Specific Output Format ──
"Write an Ansible playbook that installs nginx on RHEL 9.
Requirements:
- Enable nginx on boot
- Open port 80 in firewalld
- Deploy an index.html with content 'Hello from Ansible'
- Fix SELinux context on /var/www/html
Output: only the YAML, no explanation."

── Pattern 4: Debug with Error ──
"I'm getting this error when running terraform apply:
Error: Error creating VPC: InvalidVpcCidr.MalformedInput
My main.tf has cidr_block = '10.0.0.0/8'
What's wrong and how do I fix it?"

── Pattern 5: Explain in Simple Terms ──
"Explain Kubernetes namespaces as if I know Linux but
have never touched Kubernetes. Use a real-world analogy."

# Install GitHub Copilot extension in VS Code
# Extensions → search "GitHub Copilot" → Install

# Sign in with GitHub account
# Free for students → apply at: education.github.com
# $10/month for individuals otherwise

# How it works — as you type, Copilot suggests completions
# Accept suggestion: Tab
# Reject: Esc
# See alternatives: Alt + ]

# Copilot Chat (in VS Code sidebar)
# /explain  → explains selected code
# /fix      → suggests fix for selected code
# /tests    → generates unit tests
# /doc      → writes documentation

# pip install openai
from openai import OpenAI

client = OpenAI(api_key="sk-your-api-key-here")  # get from platform.openai.com

response = client.chat.completions.create(
    model="gpt-4o-mini",   # cheapest good model — ~$0.15 per 1M tokens
    messages=[
        {"role": "system",  "content": "You are a senior Linux sysadmin. Be concise and practical."},
        {"role": "user",    "content": "Explain this error: ENOSPC: no space left on device"}
    ]
)

print(response.choices[0].message.content)

#!/usr/bin/env python3
# log-analyser.py — paste a log, get a plain-language explanation

import sys
from openai import OpenAI

client = OpenAI(api_key="your-key")  # or use Ollama for free

def analyse_log(log_content):
    prompt = f"""You are a senior Linux/DevOps engineer.
Analyse this server log and provide:
1. What happened (plain language, 2 sentences max)
2. Root cause (if identifiable)
3. Recommended fix (specific commands if applicable)

Log:
{log_content}"""

    response = client.chat.completions.create(
        model="gpt-4o-mini",
        messages=[{"role":"user","content":prompt}]
    )
    return response.choices[0].message.content

# Read from file or stdin
if len(sys.argv) > 1:
    log = open(sys.argv[1]).read()
else:
    log = sys.stdin.read()

print(analyse_log(log))

# Simple RAG (Retrieval-Augmented Generation)
# Load your own docs → AI answers questions about them

def chat_with_docs(docs_folder, question):
    # Read all markdown/text files
    context = ""
    import os
    for fname in os.listdir(docs_folder):
        if fname.endswith(('.md', '.txt')):
            context += open(f"{docs_folder}/{fname}").read() + "\n\n"

    prompt = f"""Answer this question using ONLY the documentation below.
If the answer isn't in the docs, say so.

Documentation:
{context[:8000]}  # limit to context window

Question: {question}"""

    response = client.chat.completions.create(
        model="gpt-4o-mini",
        messages=[{"role":"user","content":prompt}]
    )
    return response.choices[0].message.content

print(chat_with_docs("./docs", "How do I configure nginx for HTTPS?"))

# Launch g4dn.xlarge on AWS (NVIDIA T4 GPU)
# AMI: Deep Learning AMI (Ubuntu) — GPU drivers pre-installed

# SSH into your EC2, then:
curl -fsSL https://ollama.com/install.sh | sh

# Pull models
ollama pull llama3.2
ollama pull codellama

# Start Ollama listening on all interfaces (for team access)
OLLAMA_HOST=0.0.0.0:11434 ollama serve &

# Open EC2 Security Group: port 11434 from your team's IPs only

# Deploy Open WebUI for browser access
docker run -d -p 80:8080 \
  -e OLLAMA_BASE_URL=http://localhost:11434 \
  -v open-webui:/app/backend/data \
  ghcr.io/open-webui/open-webui:main

# Team accesses via: http://your-ec2-ip
# Everyone gets a private AI — no data leaves your AWS account

# Cost control: Stop instance when not in use
# Use Lambda to auto-stop at night (see Cloud course Module 13)