I Published a Python Package in 2019. Revisiting It in 2026 Reminded Me of Something

Diksha Rawat — Sat, 14 Mar 2026 16:26:10 +0000

You know how elders always start with "In those days..." 😄

Well, consider this my version of that.

Because in 2019, building and publishing your own Python package wasn't something many developers around me were doing. There was no AI to guide you, no tools generating everything automatically.

You figured things out yourself. And when it finally worked, it meant something.

That little package I built back then made me genuinely proud of myself.
Revisiting it in 2026 brought a smile back to my face.

How It Started

Back in 2019, I was a few years into my career and preparing for technical interviews.

Back then, interviewers loved asking Python logic questions, not just to see if you could write code, but whether you truly understood how things worked.

While practicing problems, I noticed something interesting.

Python has packages for almost everything. But not for counting set bits in a range of natural numbers.

So I built one. 😄

What is a Set Bit?

A set bit is simply a binary digit that equals 1.

Example:

3 in binary = 011 → 2 set bits
7 in binary = 111 → 3 set bits

At this point you might be thinking, Why do we even need this? 😄

Honestly, I had the same thought later. But 2019-me probably needed it for something and decided to solve the problem.

My package counts set bits in:

a single number
or across a range of natural numbers

…with one simple function call.

Building it in 2019, No AI, Just Google

I figured everything out the hard way.

Python package structure
PyPI publishing
Documentation
Versioning

All through Google, Stack Overflow, and trial-and-error.

No ChatGPT.
No Claude.
No AI assistant.

Just curiosity, patience, and a lot of browser tabs. 😄

I wasn't a fresher building a toy project.
I was a working developer who noticed a small gap and decided to fill it.

And when the package finally went live on PyPI, I felt genuinely proud.

Maybe nobody would ever use it.

But that was never really the point.

Fast Forward to 2026

That little project stayed somewhere in the back of my mind.

So recently I went back to it.

Looking at the old code with fresh eyes, and with AI helping this time, I realised there were things I could improve:

missing input validation
no error handling
incomplete docstrings

So I cleaned it up and published v1.1.1 of:

setbitcounter-pkg 🚀

What Revisiting It Taught Me

The code I wrote in 2019 wasn’t perfect.
But it was real.

I built something from scratch, published it, and moved forward in my career.

Coming back to it in 2026 reminded me of something important.

It reminded me of the version of myself who was curious enough to explore, patient enough to learn, and motivated enough to build.

Today we have AI that can help us build almost anything.
But back then, you had to really want to figure things out.

And that curiosity still matters.

Try It Yourself

Install the package:

pip install setbitcounter-pkg

Example usage:

from setbitcounter.countset import countsetbit

print(countsetbit(7))       # 3
print(countsetbit(1, 10))   # 17

PyPI
https://pypi.org/project/setbitcounter-pkg/

GitHub
https://github.com/diksha-rawat/SetBitCount

Do you have an old project you’re proud of?
Drop it in the comments. I would love to see it! 👇

How I Used One AI to Build Another: A Local AI Code Review Pipeline (For Free)🤖

Diksha Rawat — Fri, 13 Mar 2026 16:54:42 +0000

We all hear about AI almost every minute now.

Some people are not happy about it.
Some are just letting it pass.
And some are excited to play with it.

I am in the last category. 😄

I’m fortunate enough to work in an organization that gives us access to AI models and tools to learn and build with.
But one day, sitting at home, I wondered —

What if I want to recreate something I already use at work?
Could I build it without spending money on cloud AI services?

So I did what any curious person would do —
asked one AI to teach me how to build with another AI. For free. 😄

What I Am Building

Here’s roughly how the pipeline works:

👨‍💻 Developer opens Pull Request
               ↓
⚙️  GitHub Actions triggers
               ↓
🐳  Docker builds the app
               ↓
☸️  Kubernetes deploys it
               ↓
🤖  Ollama + Mistral reviews the code diff
               ↓
💬  AI posts review comment on the PR
               ↓
👨‍💻  Developer gets feedback instantly

The interesting part is that the AI model runs locally using Ollama, so the entire system works without cloud APIs or paid services. In simple terms, it's a pipeline that automatically reviews code using a local AI model.

Is it something new? Not really. Many people have experimented with this idea already.

But my goal is different.

I want to build it from scratch, understand how it works, and share the process so others can learn as well — without needing expensive AI APIs or cloud infrastructure.

The entire system runs locally on a laptop.

No cloud.
No API costs.
No subscriptions.

Just open-source tools, your machine, and curiosity.

Who This Is For

Anyone curious about AI + DevOps and who wants to learn by building things.

If you want to explore how AI can be integrated into developer workflows without spending money, follow along. 🚀

Project: DevMind-AI

Whether you’re skeptical about AI, ignoring it, or excited to experiment like me — I’d love to hear where you stand. 😊

DEV Community: Diksha Rawat