DEV Community: shubham randive

Making your saas agentic does not mean creating a standalone chatbot with access of mcp tools and apis. It's a art to think like a product manager and integrate agents deep within the app.
shubham randive — Mon, 14 Sep 2026 04:01:32 +0000

Checkout my mcp server you can use to analyse the csv data. https://github.com/shubham303/meelu-analytics-mcp. Its not simple text-to-sql engine, but runs real statistical data science algorithms on data. I hope someone builds AI agent app on top of it.

shubham randive — Fri, 04 Sep 2026 19:02:34 +0000

GitHub - shubham303/meelu-analytics-mcp: Ask your AI assistant real questions about your data and get answers you can rely on — an MCP server with 45 analysis tools that pick the right statistical method themselves and tell you how much to trust the result. · GitHub

Ask your AI assistant real questions about your data and get answers you can rely on — an MCP server with 45 analysis tools that pick the right statistical method themselves and tell you how much t...

github.com

I open-sourced my meelu data scientist mcp 🚀 With Meelu, you can connect your CSV files to an AI agent and simply ask questions about your data. https://github.com/shubham303/meelu-analytics-mcp

shubham randive — Thu, 03 Sep 2026 10:35:21 +0000

GitHub - shubham303/meelu-analytics-mcp: Ask your AI assistant real questions about your data and get answers you can rely on — an MCP server with 45 analysis tools that pick the right statistical method themselves and tell you how much to trust the result. · GitHub

github.com

🧠 Boost Your Python Code Instantly with This LLM Knowledge Base Prompt

shubham randive — Mon, 09 Jun 2025 17:38:42 +0000

Write cleaner, more efficient code — without spending hours refactoring manually.

At the beginning of this year, I made a resolution: level up my Python skills.

I dove into classics like:

Fluent Python
Robust Python
High Performance Python
Python Cookbook
Pythonic Programming

My understanding of Python deepened. But there was a problem: even after soaking up all that knowledge, manually applying best practices to real code was tedious and slow.

So like many developers today, I turned to LLMs like ChatGPT and Claude for help.

And while they “knew” Python, their code suggestions didn’t always meet the mark:

Too much commenting that cluttered code
Over-refactoring: turning simple logic into dozens of tiny unreadable functions
Missing subtle improvements like using generators over lists

Then I Had an Idea 💡

Claude introduced a new feature: Claude Projects — letting you upload files the LLM can reference like a knowledge base.

So I built one.

I gathered everything I’d learned:

Best practices
Memory & performance tips
Idiomatic patterns
Anti-patterns
“Good vs Bad” examples

I created a single markdown file that acts as a comprehensive Python knowledge base.

And then… I gave it to the LLM.

The difference was night and day.

Here’s a Real Example

My Original Code:

def process_file(filename):
    with open(filename, 'r') as file:
        lines = file.readlines()

    for i in range(len(lines)):
        lines[i] = lines[i].upper()

    with open(filename, 'w') as file:
        file.writelines(lines)

LLM’s Refactored Code (with Knowledge Base Prompt):

import tempfile
import shutil
import logging
from pathlib import Path

log = logging.getLogger(__name__)

def convert_file_to_uppercase(filename: str) -> bool:
    """Convert file contents to uppercase using temporary file for safety."""
    file_path = Path(filename)

    try:
        with tempfile.NamedTemporaryFile(
            mode='w',
            delete=False,
            dir=file_path.parent,
            prefix=f".{file_path.name}_temp_",
            suffix=".tmp"
        ) as temp_file:
            temp_filename = temp_file.name

            with open(filename, 'r') as input_file:
                for line in input_file:
                    temp_file.write(line.upper())

        shutil.move(temp_filename, filename)
        log.info("file.converted", filename=filename)
        return True

    except FileNotFoundError:
        log.error("file.not_found", filename=filename)
        return False
    except PermissionError:
        log.error("file.permission_denied", filename=filename)
        return False
    except OSError as e:
        log.error("file.operation_failed", filename=filename, error=str(e))
        try:
            Path(temp_filename).unlink(missing_ok=True)
        except (NameError, OSError):
            pass
        return False

if __name__ == "__main__":
    logging.basicConfig(level=logging.INFO)
    success = convert_file_to_uppercase('your_file.txt')
    result_msg = "File successfully converted" if success else "Failed to process file"
    log.info("conversion.complete", success=success, message=result_msg)

What Got Better?

✅ Safer file handling with tempfile
✅ Type annotations
✅ Clear function name and docstring
✅ Memory-efficient line-by-line processing
✅ Exception handling
✅ Logging

And it all came from just one prompt.

What’s Inside the Prompt?

This prompt isn't just a file — it's a developer’s assistant.

Here’s what’s included:

✅ Python best practices across readability, safety, and performance
✅ “Bad vs Good” code examples for every pattern
✅ Idiomatic Python tricks (like with, enumerate, defaultdict, etc.)
✅ Clear advice on error handling, logging, and type hinting
✅ LLM-friendly format — works with Claude, ChatGPT, or any model that supports file uploads or custom prompts

How to Use It

Option 1: Claude Projects

Create a project (e.g., “Python Best Practices”)
Upload the markdown knowledge base
Paste the system_prompt.txt file as the project prompt
Ask Claude to refactor or review your code

Option 2: ChatGPT Custom GPT (for Plus users)

Create a new GPT
Upload the same files
Now when you provide your Python code, ChatGPT will use the uploaded best practices to suggest improvements

Option 3: Manual Upload (Any LLM)

Upload the markdown + system prompt directly in a conversation, then paste your code.

Try It Out

We’ve created two versions:

🎁 Free Version

Covers key tips
Includes several good/bad code examples
Try it before you buy it

💼 Full Version — \$2

Full best practices guide
Regular updates
One-time payment

👉 Get it on Gumroad

Who Should Use This?

🧑‍💻 Intermediate to advanced Python developers
⚙️ Engineers who use LLMs in their workflow
⏱ Anyone who wants to save hours of time cleaning up and optimizing code

Final Thoughts

Tools like ChatGPT and Claude are powerful. But they work best when you feed them high-quality, domain-specific context.

This knowledge base prompt turns your favorite LLM into a Python best practices expert — one that never gets tired and is always up to date.

Start writing better Python code today.

How I Earned $1,000 as a Python Developer Without Moonlighting

shubham randive — Sat, 07 Jun 2025 14:45:33 +0000

How I Turned My Python Pain Points Into a $1K Side Product

A story about building smarter AI coding assistants and accidentally creating a business

The Problem Every Python Developer Faces

I'm a Python developer who takes pride in writing clean, efficient code. You probably are too.

But here's what frustrated me: Even after reading Fluent Python, Robust Python, High Performance Python, and other classics in our field, I kept making the same small, avoidable mistakes in my daily work.

It wasn't a knowledge problem—it was a memory problem. Performance optimizations, naming conventions, idiomatic patterns... when you're deep in the flow of building something, it's easy to fall back into old habits.

That's when I had an idea: What if I could create a tool that remembers all the best practices for me?

Why Generic AI Wasn't Cutting It

My first approach was simple: paste my code into ChatGPT or Claude and ask for improvements.

The results were... inconsistent:

They missed subtle performance issues
Sometimes gave over-engineered solutions (hello, unnecessary single-responsibility principle violations)
Couldn't proactively catch issues unless I asked the exact right question

What I needed was an AI assistant that already knew the specific best practices I cared about.

Building My Python Knowledge Base

So I went all-in on creating my own solution.

Over several weeks, I compiled a comprehensive markdown file containing:

All major Python topics (functions, OOP, iterators, memory management)
Best practices from every Python book I'd read
Code examples showing good vs bad patterns
Common refactoring techniques and performance tweaks
Idiomatic Python tricks and conventions

I kept refining examples, adding real-world use cases, and polishing the content until I had something substantial.

Then I plugged it into Claude Projects and Custom GPTs with a tailored system prompt.

The results? Game-changing.

Real Example: From Rookie to Production-Ready

Here's a function I had written for a quick file processing task:

def process_file(filename):
    with open(filename, 'r') as file:
        lines = file.readlines()

    for i in range(len(lines)):
        lines[i] = lines[i].upper()

    with open(filename, 'w') as file:
        file.writelines(lines)

Here's what my enhanced AI assistant suggested:

import tempfile
import shutil
from pathlib import Path

def convert_file_to_uppercase(filename: str) -> bool:
    """Convert file contents to uppercase using temporary file for safety."""
    file_path = Path(filename)

    if not file_path.exists() or not file_path.is_file():
        print(f"Error: Invalid file '{filename}'")
        return False

    try:
        with tempfile.NamedTemporaryFile(mode='w', delete=False,
                                         dir=file_path.parent,
                                         prefix=f".{file_path.name}_temp_",
                                         suffix=".tmp") as temp_file:

            with open(filename, 'r') as input_file:
                for line in input_file:
                    temp_file.write(line.upper())

        shutil.move(temp_file.name, filename)
        return True
    except OSError as e:
        print(f"Error processing file: {e}")
        return False

What improved:

Descriptive function naming
Type annotations for clarity
Robust file validation
Memory-efficient line-by-line processing
Safe file operations using temporary files
Proper error handling and logging
Clear return values for success/failure

All from a single request. No back-and-forth, no "can you make this more pythonic?" follow-ups.

Turning Pain Into Profit

After months of using this system and seeing dramatic improvements in my code quality, I realized other developers might want this too.

So I packaged it as a product on Gumroad:

💡 Free tier: Core topics, basic examples
📘 $2 one-time: Full content with comprehensive good/bad code examples
🚀 $5/year: Everything plus lifetime updates

The Results

Without writing blog posts or running ads, the product found its audience through:

Developer forums
AI/LLM communities
Word of mouth from early users

Within a few months: $1,000+ in sales

No client work, no contract gigs, no complex marketing funnels—just solving a problem I had and sharing the solution.

Want to Try It?

If you're tired of your AI giving generic Python advice, you can check out my knowledge base here:

👉 One-time purchase

👉 Annual subscription

There's a free version if you want to test it out first.

Key Takeaways

This experience taught me three valuable lessons:

You don't need to be a tech influencer to build something profitable
Your biggest frustrations often point to the best product opportunities
AI tools are powerful, but they need the right context to be truly useful

If you've built internal tools or systems that make your development life easier, consider whether other developers might benefit from them too.

Sometimes the best products come from solving your own problems first.

What's your biggest Python pain point? Let me know in the comments—maybe it's the next product opportunity waiting to be discovered.

DEV Community: shubham randive

Making your saas agentic does not mean creating a standalone chatbot with access of mcp tools and apis. It's a art to think like a product manager and integrate agents deep within the app. shubham randive — Mon, 14 Sep 2026 04:01:32 +0000

Checkout my mcp server you can use to analyse the csv data. https://github.com/shubham303/meelu-analytics-mcp. Its not simple text-to-sql engine, but runs real statistical data science algorithms on data. I hope someone builds AI agent app on top of it.

GitHub - shubham303/meelu-analytics-mcp: Ask your AI assistant real questions about your data and get answers you can rely on — an MCP server with 45 analysis tools that pick the right statistical method themselves and tell you how much to trust the result. · GitHub

I open-sourced my meelu data scientist mcp 🚀 With Meelu, you can connect your CSV files to an AI agent and simply ask questions about your data. https://github.com/shubham303/meelu-analytics-mcp

GitHub - shubham303/meelu-analytics-mcp: Ask your AI assistant real questions about your data and get answers you can rely on — an MCP server with 45 analysis tools that pick the right statistical method themselves and tell you how much to trust the result. · GitHub

🧠 Boost Your Python Code Instantly with This LLM Knowledge Base Prompt

Then I Had an Idea 💡

Here’s a Real Example

My Original Code:

LLM’s Refactored Code (with Knowledge Base Prompt):

What Got Better?

What’s Inside the Prompt?

How to Use It

Option 1: Claude Projects

Option 2: ChatGPT Custom GPT (for Plus users)

Option 3: Manual Upload (Any LLM)

Try It Out

🎁 Free Version

💼 Full Version — \$2

Who Should Use This?

Final Thoughts

How I Earned $1,000 as a Python Developer Without Moonlighting

How I Turned My Python Pain Points Into a $1K Side Product

The Problem Every Python Developer Faces

Why Generic AI Wasn't Cutting It

Building My Python Knowledge Base

Real Example: From Rookie to Production-Ready

Turning Pain Into Profit

The Results

Want to Try It?

Key Takeaways

Tags

Making your saas agentic does not mean creating a standalone chatbot with access of mcp tools and apis. It's a art to think like a product manager and integrate agents deep within the app.
shubham randive — Mon, 14 Sep 2026 04:01:32 +0000