My first AI experince

Author: Ricky Khatri

 

My Journey Into AI as a Software Engineer

When ChatGPT became popular, I had the same reaction many software engineers did.

I was curious, a little skeptical, and honestly wondering whether it was just another tech trend that would disappear after the hype.

I was wrong.

Over the past few months, I've spent time building applications with AI instead of just reading about it. I experimented with language models, prompt engineering, retrieval-augmented generation (RAG), AI agents, vector databases, and API integrations. Some ideas worked surprisingly well. Others failed completely.

That experience changed how I think about software engineering.

AI Doesn't Replace Software Engineers

One of the biggest misconceptions is that AI replaces developers.

In reality, AI writes code, but software engineering is much more than writing code.

Someone still has to:

• Design scalable architectures

• Understand business requirements

• Handle edge cases

• Secure applications

• Optimize performance

• Review generated code

• Deploy and monitor production systems

AI accelerates development, but engineers are still responsible for building reliable software.

Think of AI as an incredibly fast junior developer who never gets tired but still needs guidance.

My Biggest Lesson

I quickly realized that prompts are not the product.

The real challenge is building systems around AI.

For example:

A chatbot isn't just a prompt.

It needs authentication, APIs, databases, caching, logging, monitoring, security, rate limiting, testing, and deployment.

That's where software engineering becomes even more valuable.

AI Makes Me Faster

Today, I use AI for doing 30% of my work.

Not to replace my thinking, but to accelerate it.

I use it for:

• Exploring new technologies

• Validating UI/UX designs for responsiveness

• Writing unit tests

• Refactoring legacy code

• Explaining unfamiliar libraries

• Reviewing implementation ideas

• Drafting documentation

• Code review and commenting

Instead of spending hours on repetitive work, I can focus more on architecture and solving business problems.

What Still Requires Human Thinking

AI can generate code.

It cannot fully understand your users, your business, or your team's constraints.

It doesn't know why a feature exists.

It doesn't know what happened in that production outage last month.

It doesn't know the trade-offs your company made two years ago.

Good software engineering is still about making informed decisions under uncertainty.

That's something experience teaches.

Advice for Developers Starting with AI

If you're beginning your AI journey, don't try to learn everything at once.

Start small.

Build a simple chatbot.

Call an LLM API.

Experiment with embeddings.

Create a RAG application.

Deploy it.

Break it.

Fix it.

You'll learn far more by building than by watching tutorials.

Looking Ahead

I believe AI will become a standard part of every developer's toolkit, much like Git, Docker, or cloud platforms.

The engineers who thrive won't be the ones who fear AI.

They'll be the ones who understand how to combine software engineering principles with AI capabilities to build better products.

This blog marks the beginning of my journey in sharing what I learn while building AI-powered applications.

I don't claim to know everything.

But I do believe the best way to learn is to build, share, and keep improving.

Thanks for reading. More technical deep dives, architecture discussions, and hands-on AI projects are coming soon.