How to Be a Developer in a Post-AI World

Aman Jaiswal
5 min read5 days ago

I recently explored what it feels like to write code with AI and accomplish tasks without writing a single line of code. Here’s what I experienced, along with some predictions on where we’re headed this year.

Talking about predicting the future

Back in ’19, when I first joined tech, we had a three-month bootcamp where I learned everything needed to get started: Git versioning, the basics of web development, Bootstrap + CSS + HTML, using frameworks like Flask & Django, writing tests first and code later, and many other skills.

I remember Corey Schafer teaching me how to write production level code and actually deploy it for free, so it runs on a URL other than localhost.

I remember going through series after series, trying to understand why writing tests is important, what patching and mocking are, and everything in between.

There’s one guy who taught everything. From Machine Learning to Go, to Data Analysis, to how to be curious about things.

These folks are some of the heroes I grew up with. After spending a few years in this field, I realized there’s so much more I can learn if I just read a little more. Some of the amazing books I’ve read have taught me how to write better, think a little harder, and approach writing code the way one tells a story. I’m certain everyone has at least seen the covers, If not read. Books like:

  1. Clean Code
  2. The Pragmatic Programmer

and many more like them.

While working on frontend, someone once forwarded me an article about why you should write fewer unit tests. It sounded like a crazy idea, but a valid one nonetheless.

In ’25, just a few days ago, with just a handful of prompts, I was able to achieve close to 50% code coverage. Working with tools like Cline and Cursor really makes you realise — it’s incredible to accomplish in a few hours what used to take weeks.

The process can be painful at times because there’s a high chance things will go wrong unless you’re explicitly clear about what you want to accomplish. But nonetheless, with enough attempts and a large enough context window, you can get back on track, even if you don’t fully understand the nuances of how they work, like attention mechanisms, chain of thought, transformers, and so on.

We didn’t have this superpower before. There were countless times in my career when I would pray just to find a single answer that could help me debug a massive error stack trace — one twice the length of my actual code. You don’t need to do that kind of praying anymore. Now, you can simply put more and more agents to work, and eventually, one of them will come back with the right answer.

I feel that as humans, we’ve grown by solving problems — big or small.

For example, Dependency Hell.

Finding the exact versions of dependencies that play nice with each other used to take hours whenever you upgraded your tech stack. Now, it takes minutes.

Today, you have the entire internet, trained and just a few clicks away, ready to solve all your problems.

Today, you have the entire internet, trained and just a few clicks away, ready to solve or guide you through almost anything. So you make fewer mistakes, way fewer. Writing an end to end web app took me 30 minutes, something that I had done before in more than a week when I was an intern.

Sadly, The joy seems distant now. You can instruct and get everything done, but the language you instruct in is now mostly English.

You can write poems,
You can learn philosophy,
You can ship code faster than any human ever before,
You can do it all with close to 100% accuracy most of the time.

Yet, you might no longer be an expert in any of them — at least not line by line. But you can supervise, if that still counts.

What’s next?

The need of the hour is different now. It’s about expanding how you think, how you see things, and how much you can predict correctly.

The tools are here, free and at your service. The real question is: what do you want to build?

Even though I’ve worked across frontend, backend, data, and machine learning over the past six years, I believe now is the time for everyone to adopt a product mindset.

We’ll talk more in metrics than in code quality.
We’ll focus more on how quickly something can be shipped and less on how stable it will remain.

There are no “best practices” anymore — just coding standards that everyone follows.

The age of the hackers is over. Now, it’s time for product thinkers and builders to rise and create faster than ever before.

If there ever were a Renaissance in the 21st century, this would be it. Thinkers will be valued again, maybe even more than doers at times.

So the choice is simple. It’s not about whether you should adopt this mindset and learn the necessary skills. It’s about when.

In other things, Some of the great content I experienced in past few weeks

A Real Pain (Romulus after he was in Succession)
The Darjeeling Limited (before Adrian Broody’s second holocaust survival)
The Shadow Strays (the guy who did The Night Comes For Us)
Between Flowers (OAFF, in his own skin)
Interstellar (damn that docking scene)
Pantheon (nothing comes close to predicting a world with multiple AI and frontier models)
Gravity Falls (just the theme song maybe)

fin.

Free

Distraction-free reading. No ads.

Organize your knowledge with lists and highlights.

Tell your story. Find your audience.

Membership

Read member-only stories

Support writers you read most

Earn money for your writing

Listen to audio narrations

Read offline with the Medium app

Aman Jaiswal
Aman Jaiswal

Written by Aman Jaiswal

A part time nerd and a full time engineer. Loves to talk about tech, cinema and everything in between.

No responses yet

Write a response