From Curiosity to Code: My Journey Into Software Development

It Started With Curiosity

I didn't grow up around computers or have a CS degree handed to me. I came from Rourkela, Odisha a steel city where engineering usually meant metallurgy, not software. But somewhere between school and figuring out what to do next, I stumbled into programming.

I remember the first time I wrote a piece of code that actually did something. It wasn't elegant. It probably had ten bugs. But watching the screen respond to something I typed that was the moment I knew this was what I wanted to do.

Learning in Public

I started with Java the classic DSA grind that every self-taught developer knows too well. Arrays, linked lists, trees, graphs. Hours of solving problems that felt impossible until they suddenly clicked.

Then I discovered web development, and everything changed. HTML and CSS felt like building with Lego. JavaScript made those blocks come alive. React made me feel like I could build anything.

But the real turning point was going freelance. Nothing teaches you faster than a client waiting for a deliverable. I learned to build responsive websites, integrate APIs, handle payments, and most importantly communicate with people who don't speak code.

The Leap to Production

In October 2024, I joined BigCircle as a Software Developer. Going from freelance projects to production enterprise systems was a different world entirely.

Suddenly I was working with:

• tRPC for type-safe APIs that catch bugs before they reach production
• FastAPI and Node.js backends that need to handle real traffic
• Docker and AWS for deployments that can't go down
• LLM routing and observability building AI systems that actually work reliably

The biggest shift wasn't technical it was learning to build systems that other people depend on. When your code runs in production, "it works on my machine" isn't good enough.

Building Things That Matter

The projects I'm most proud of aren't the ones with the fanciest tech stack. They're the ones that solved real problems.

Ubik, an AI desktop assistant, taught me how to orchestrate multiple AI agents to handle complex tasks like email management. Building it meant understanding not just the code, but how people actually work with their inboxes.

An AI document classification system showed me that the gap between a demo and a production system is massive. OCR, webhooks, background processing with Redis, fallback strategies real-world AI is 20% model and 80% engineering.

What I've Learned So Far

A few honest lessons from the road:

1. The best code is the code you don't write. Every line is a liability. Simplicity wins.
2. Learn by building, not by watching tutorials. Tutorials give you confidence. Building gives you competence.
3. Your first version will be bad. Ship it anyway. I've never looked at old code and thought "wow, that was perfect." Growth requires cringing at your past work.
4. The tech stack matters less than you think. Frameworks come and go. Problem-solving skills don't.
5. Ask for help. The developer community is incredibly generous. Don't suffer in silence.

What's Next

I'm currently pursuing my BCA from IGNOU while working full-time balancing formal education with hands-on experience. I believe both matter.

Going forward, I want to write more about the things I'm learning in the trenches: building AI systems that work in production, fullstack patterns that scale, and the tools that make development less painful.

This blog is my space to share those lessons no fluff, no hype, just real stuff from a developer who's still figuring it out.

Thanks for reading. Let's build something.