DEV Community: Samarth Funde

AI_Assisted_Devops_Day-3_Project-3_DeepSeek_LLM

Samarth Funde — Mon, 08 Dec 2025 11:37:35 +0000

Generate_Dockerfile_using_Ollama_Hosted(DeepSeek)_LLM project-3

Required:
Ollama → runtime to manage and run the model.
LLaMA / deepseek model → the actual AI model files.

Not required:
Python / virtual environment → only needed if you want to run Python scripts that call the model programmatically (like your Dockerfile generator script).

Step A) Setup ollama in your local EC2 server

Step 1) Create EC2 instance with enough Storage and Ram instance i have used (t3.medium) and ubuntu OS

Step 2) go to ollama.com website search llama3 and select the version llama3.2:1b

Step 3) you can see in website Download button click on it and you see there Linux so copy following commands and paste in your instace

a) Download & Install Ollama for linux:
curl -fsSL https://ollama.com/install.sh | sh
b) Start ollama service: ollama serve or start &
c) Run the model: ollama run llama3.2:1b
<<.......... now you can write here your query e.g 'Create a Dockerfile based on Java application'
....................................................................................................................................................

Step B) go on ollama.com website and find the model of 'DeepSeek' choose the version 'deepseek-r1:1.5b'
Step 4) run this command: ollama run deepseek-r1

....................................................................................................................................................

if you want to visual view for your ai model then you can download Chatbox ..... and here we can use like local model

AI_Assisted_Devops_Day-3_Project-2_Gemini_LLM

Samarth Funde — Mon, 08 Dec 2025 10:15:26 +0000

Generate_Dockerfile_using_Ollama_Hosted(Gemini)_LLM project-2

Step A) Setup ollama in your local EC2 server

Step 1) Create EC2 instance with enough Storage and Ram instance i have used (t3.medium) and ubuntu OS

Step 2) go to ollama.com website search llama3 and select the version llama3.2:1b

Step 3) you can see in website Download button click on it and you see there Linux so copy following commands and paste in your instace

............................................................................................................................................

Step B) Setup Python Virtual Environment in EC2 instance

Create Virtual Environment copy & paste the following cmds
sudo apt update
sudo apt install python3 python3-venv python3-pip -y
python3 -m venv venv
source venv/bin/activate # On Linux/MacOS

....................................................................................................................................................

Step C) generate the API Key from AI Google Studio

Step 5) find the API Key button and generate the API key for gemini

....................................................................................................................................................

Step D) Create a Python files for run the Local-LLM

Step 6) echo "google.generativeai" > requirement.py
install dependancies:
pip3 install -r requirement.txt

Step 7) nano generate_dockerfile.py ...... and copy paste code

import os
import google.generativeai as genai
import ollama

Set your API key here
os.environ["GOOGLE_API_KEY"] = "XXXXXXXXXXXXXXXXXXX"

Configure Gemini Model
genai.configure(api_key=os.getenv("GOOGLE_API_KEY"))
model = genai.GenerativeModel('gemini-1.5-pro')

PROMPT = """
Generate an ideal Dockerfile for {language} with best practices.

Include:

Base image
Installing dependencies
Setting WORKDIR
Adding application source code
Multi-stage build (builder + production)
Command to run the application
"""
def generate_dockerfile(language):
response = ollama.chat(
model='llama3.2:1b',
messages=[{'role': 'user', 'content': PROMPT.format(language=language)}]
)

Correct response key
return response['message']['content']

if name == 'main':
language = input("Enter the programming language: ")
dockerfile = generate_dockerfile(language)
print("\nGenerated Dockerfile:\n")
print(dockerfile)

Step 8) after run above file you can see output 'Enter the programming language: node.js' so write as per your required language eg, java, groovy, python,nodejs etc.. it will generate the Dockerfile

AI_Assisted_Devops_Day-3_Project-1_Local_LLM

Samarth Funde — Mon, 08 Dec 2025 06:04:47 +0000

Generate_Dockerfile_using_Ollama_Local_LLM project-1

Step A) Setup ollama in your local EC2 server

Step 1) Create EC2 instance with enough Storage and Ram instance i have used (t3.medium) and ubuntu OS

Step 2) go to ollama.com website search llama3 and select the version llama3.2

Step 3) you can see in website Download button click on it and you see there Linux so copy following commands and paste in your instace
a) Download & Install Ollama for linux:
curl -fsSL https://ollama.com/install.sh | sh
b) Start ollama service: ollama serve or start &
c) Run the model: ollama run llama3.2:1b
<<.......... now you can write here your query e.g 'Create a Dockerfile based on Java application'

.......................................................................

Step B) Setup Python Virtual Environment in EC2 instance

Step 4) Create Virtual Environment copy & paste the following cmds
sudo apt update
sudo apt install python3 python3-venv python3-pip -y
python3 -m venv venv
source venv/bin/activate # On Linux/MacOS

........................................................................

Step C) Create a Python files for run the Local-LLM

Step 5) echo "ollama" > requirement.py
install dependancies:
pip3 install -r requirement.txt

Step 7) nano generate_dockerfile.py ...... and copy paste code

import ollama
PROMPT = """
ONLY Generate an ideal Dockerfile for {language} with best practices. Do not provide any description
Include:

Base image
Installing dependencies
Setting working directory
Adding source code
Running the application
Multi stage docker file """

def generate_dockerfile(language):
response = ollama.chat(
model='llama3.2:1b',
messages=[{'role': 'user', 'content': PROMPT.format(language=language)}]
)
return response['message']['content']

if name == 'main':
language = input("Enter the programming language: ")
dockerfile = generate_dockerfile(language)
print("\nGenerated Dockerfile:\n")
print(dockerfile)

Run the Application:
python3 generate_dockerfile.py

Step 8) after run above file you can see output 'Enter the programming language: java' so write as per your required language eg, java, groovy, python,nodejs etc.. it will generate the Dockerfile

Step 9) 'Enter the programming language: groovy'

AI_ASSISTED_DAY_2 (Prompt Engineering)

Samarth Funde — Fri, 05 Dec 2025 08:00:45 +0000

✅ What is Prompt Engineering?

Prompt engineering is the process of designing and optimizing prompts (instructions given to AI models like ChatGPT) so that the model gives accurate, useful, and predictable outputs.

It focuses on:
How to ask
What context to include
How to structure the input

Reducing token usage (cost optimization)

✅ Cost Optimization in Prompt Engineering

Prompt engineering helps reduce token cost (usage cost) by:

1. Reducing Prompt Length
Use short and clear instructions instead of long paragraphs.

2. Using Structured Prompts
Tables, bullet points, or templates reduce repetition.

3. Reusing Context Instead of Rewriting
Give only the required information—avoid sending full documents repeatedly.

4. Caching Responses
Save earlier outputs instead of regenerating.

5. Using the Smallest Model Needed
Use cheaper/smaller models for simpler tasks.

6. Avoid Over-specifying
Don’t add unnecessary details (“write like Shakespeare”).
More tokens = more cost.

Types Of Prompting
**
**✅ Zero-Shot Prompting

Definition:
The model is asked to perform a task without giving any example.

Example:
“Classify this message as spam or not spam:
‘You won a free iPhone, click here.’”

✅ Few-Shot Prompting

Definition:
You give the model a few examples (2–5) before asking it to answer.

Example:

Example 1: Positive

Example 2: Negative

Example 3: Positive

Now classify this: "The movie was terrible."

✅ Multi-Shot Prompting

Definition:
You give the model many examples (more than a few) to teach the pattern.

Used in:

Large instructions
Training-like scenarios
Complex reasoning tasks

✅ CoT Prompting (Chain-of-Thought)

Definition:
You ask the model to show step-by-step reasoning before giving the final answer.

Example:
“Explain your steps and then answer:
If a train leaves at 3 PM and travels 2 hours, when will it arrive?”

Why it is useful:
Improves reasoning accuracy
Helps with math, logic, multi-step problems

AI_ASSISTED_DEVOPS DAY-0

Samarth Funde — Thu, 04 Dec 2025 14:06:57 +0000

Day 0

🧠 1. Traditional AI (Rule-Based AI)

✔️ Definition
Traditional AI refers to systems that follow fixed rules, logic, and pre-programmed instructions.
They cannot learn from data automatically; you must write the rules manually.

✔️ Example
“If user enters wrong password 3 times → lock the account.”
ATM machine logic
Traffic signal control
Chess engine rules
✔️ Simple Explanation

Traditional AI = Hard rules written by humans
No creativity, no learning.

🤖 2. Generative AI (GenAI)

✔️ Definition
Generative AI is a type of artificial intelligence that can create new content such as text, images, music, code, videos, etc., based on patterns learned from data.

✔️ Example
ChatGPT creating answers
Midjourney creating images
GitHub Copilot writing code
DALL·E generating photos

✔️ Simple Explanation
Generative AI = AI that creates.
It learns from large datasets and produces original content.

📚 3. LLM (Large Language Model)

✔️ Definition
An LLM is a Generative AI model trained on huge amounts of text so it can understand and generate natural language like a human.

It can:
answer questions
write code
summarize text
translate languages
chat naturally

✔️ Examples of LLMs

GPT-4, GPT-5 (OpenAI)
Claude 3 (Anthropic)
Gemini (Google)
LLaMA (Meta)

✔️ Simple Explanation
LLM = a smart AI that understands and writes language like humans.

🤖 4. What is an AI Agent?

✔️ Definition
An AI Agent is an AI system that can think, decide, and take actions automatically to complete a task using tools or external systems.

It doesn’t just chat —
it performs actions.

✔️ What an agent can do?

check system health
write and run code
call an API
read files
solve tasks step-by-step
automation (DevOps, IT operations, HR tasks)

✔️ Simple Explanation

AI Agent = AI that acts like a worker.
It can plan a task and execute steps on its own.

💬 5. Examples of Chatbots & Assistants

✔️ Chatbots (Basic)

These follow rules or basic AI:
Website customer support bot
Banking chatbot (balance, mini statement)
Swiggy/Zomato bot
Airline inquiry bot

✔️ AI Assistants (Advanced)

These use Generative AI or LLMs:
ChatGPT
Google Gemini Assistant
Microsoft Copilot
Alexa / Siri
GitHub Copilot (coding assistant)

✔️ Simple Explanation

Chatbot = answers questions
Assistant = helps you complete tasks

AI_ASSISTED_DEVOPS DAY-1

Samarth Funde — Thu, 04 Dec 2025 13:57:09 +0000

## Day1

Task : check the ec2 cpu optimization, memory optimization, disk optimization is healhty or not

Solution using AI Copilot:

Step 1) go in Github Account Create a Repositoy like 'Ai_Assisted_Devops_day1'

Step 2) Open Vs Code 'Install Copilot Extension Chat' open the terminal and write the command 'git clone githubrepourl'

Step 3) in the VS code just click the button 'ctrl+shift+i' you can see copilot is open

Step 4) in the copilot chat write this prompt *'Create a shell script where the script should analyze the health of the virtual machine based on CPU, memory, and disk space. If any of these three things are less than 60 percent utilized, the script declares the state of the VM as healthy. Whereas if any of these parameters are more than 60 percent, the script declares the health as not healthy. Also, the script should support a command-line argument named "explain". When passed, the script should explain the reason for the health status along with printing the health status. Target virtual machines are always ubuntu' *

Step 5) you can see shell file is created and all process output is generated

Step 6) you can check file code otherwise see the left menu and click on 'Source Control' icon and you can see the Commit button so write a some commit message and click on 'Commit' button

Step 7) go in Github Account and refresh the page and you get your shell file inside the repository so copy that repo code

Step 8) open the AWS Console and launch the EC2 instance

Step 9) login your instance with terminal or click on connect button.... and make a folder like 'Ai_Assited_Devops_Day1' and write the command inside this folder 'git clone github-repourl'

Step 10) and give the permission to the shell file for the execution so write the command 'chmod +x shellfilename' and './shellfilename'
and here you can see your output 'Healthy or Not'

.........................................................................................................

🐧 Learn Linux & Shell Scripting — My Free YouTube Tutorials

Samarth Funde — Wed, 03 Dec 2025 10:25:37 +0000

If you are starting your journey in Linux, Shell Scripting, Cloud, or DevOps, I create simple and beginner-friendly videos to help you learn faster.

👉 My YouTube Channel: www.youtube.com/@Cloud_with_Sam

⭐ What I Teach

Linux basics

Important daily commands
Users, permissions, file system
Shell scripting step-by-step
Practical demos on terminal
All videos are short, clear, and easy to understand.

🎯 Why Watch My Channel?

Simple explanations
Hands-on examples
Perfect for beginners
Helpful for DevOps, Cloud, and IT jobs

🚀 Start Learning Today
Subscribe to my channel and begin your Linux & Shell scripting journey:

👉 www.youtube.com/@Cloud_with_Sam

☁️ What is Cloud Computing?

Samarth Funde — Wed, 03 Dec 2025 10:17:07 +0000

Cloud means using someone else’s computers (AWS, Azure, GCP) through the internet.
Instead of buying your own servers, you can:

Create a virtual server in minutes
Store your files securely
Run applications globally
Scale up or down anytime

👍 Benefits of Cloud
No need to buy hardware
Pay only for what you use
Very fast deployment
High security and backups

🧑‍💻 Examples of Cloud Services
EC2 → Virtual server
S3 → File storage
RDS → Database
Lambda → Serverless functions