DEV Community: IamAdhitya

I Built 7 AI Apps in College — Here's What Nobody Tells You

IamAdhitya — Tue, 02 Jun 2026 14:46:47 +0000

Most students are still figuring out their first to-do app tutorial when they enter college. I decided to build a company instead.

I'm Adhitya, a B.Tech student at IITRAM, Ahmedabad. In my first two years of college, I built and shipped 7 live, monetized AI applications under my indie studio Rewrite Labs — plus 7 open source libraries used by other developers.

No CS degree required. No internship at a big tech company. Just a laptop, free tiers of incredible tools, and an obsession with shipping.

Here's everything I wish someone had told me before I started.

1. The Stack That Actually Works for Solo AI App Builders

Before I talk about lessons, here's the exact stack I use across all my apps. I landed on this after painful trial and error and I haven't needed to change it since:

Layer	Tool	Why
Frontend	React + Vite + Tailwind CSS	Fast dev, huge ecosystem
Auth	Clerk	10-minute auth setup, handles JWTs
Database	Supabase	Postgres with a clean dashboard, free tier is generous
AI	Groq API	Blazing fast inference, free tier, llama-3.3-70b is incredible
Payments	Razorpay	Best option for India, subscription plans supported
Deployment	Vercel	One command deploy, serverless functions built in

This stack lets me go from idea to live, paying app in under a week. Every tool has a free tier. Every tool has great documentation. There's no reason to use anything more complex when you're building solo.

2. Groq Will Change How You Think About AI Apps

Most people default to OpenAI. I did too, for about five minutes.

Then I tried Groq — and it felt like switching from dial-up to fiber.

Groq runs LLM inference on custom hardware called LPUs (Language Processing Units). The result? Responses that stream so fast they feel instant. For a chat app like my AI companion Nura, this is the difference between a product that feels alive and one that feels like it's thinking too hard.

The API is almost identical to OpenAI's, so migration is trivial. And the free tier is genuinely usable — not just a "taste."

If you're building any AI app today, start with Groq. You can always switch models later.

3. Ship Ugly. Polish Later. Seriously.

My first app took way too long because I kept polishing before shipping.

The brutal truth: nobody is looking at your app until you tell them to. That means every hour you spend pixel-perfecting before your first 10 users is an hour wasted.

My current process:

Build the core feature loop in 2–3 days
Deploy it live (even if it looks rough)
Get 5 real people to use it
Fix what breaks
Then polish

The version of your app you ship first will look nothing like version 1.0. Stop treating the MVP like a portfolio piece.

4. Auth is Not Your Problem — Stop Building It

Early on I wasted two days building a custom authentication system. Session tokens, password hashing, email verification — the works.

Then I discovered Clerk and deleted all of it.

Clerk gives you:

Email + social login
JWT verification for your API routes
Pre-built React components
User management dashboard

Setup takes literally 10 minutes. It's free for the first 10,000 monthly active users. There is zero scenario where you should be hand-rolling auth as a solo builder in 2025.

Same principle applies everywhere — use Supabase instead of building your own database layer, use Vercel instead of managing your own server. Your job is to build the product, not the infrastructure.

5. Monetization Is Easier Than You Think (If You Start Early)

I put a paywall in my second app. It felt bold, almost arrogant — who would pay for something a college student built?

Turns out: people who find it valuable.

Here's what I've learned about pricing AI apps as a student:

Start charging from day one. A free plan is fine, but have a paid tier. It forces you to build something worth paying for.
Subscriptions beat one-time payments. Recurring revenue means you can predict growth. Razorpay makes this straightforward with their subscription plans API.
The paywall is actually a feature. It filters out users who aren't serious. Your paying users give better feedback and stick around longer.

Don't wait until your app is "ready" to think about money. The business model should be part of the design from the start.

6. Open Source Is Your Best Marketing

While my apps are private (protecting the IP), I extracted reusable pieces into 7 open source libraries — things like:

react-premium-gate — Razorpay subscription paywall components
react-macro-rings — animated SVG nutrition progress rings
react-toast-native — lightweight toast notifications
groq-chain — Python LLM chaining without LangChain
llm-router — routes prompts to the right LLM by complexity and cost

Each library is a tutorial disguised as code. Developers find them, use them, and discover who built them. It's the most authentic form of marketing there is — you're not selling anything, you're just being useful.

If you're building apps, look for the pieces you keep rebuilding and open source them. It builds your GitHub profile, your reputation, and your network simultaneously.

7. The Thing Nobody Tells You About Building in College

Everyone focuses on the technical side. The harder part nobody talks about is the mental game.

You'll have weeks where nothing works. Where a bug takes three days to find and it's a single missing await. Where you ship something and nobody cares. Where your college workload collides with a critical bug in production.

What keeps you going isn't motivation — motivation is unreliable. What keeps you going is having built the habit of showing up and shipping anyway.

Some things that helped me:

Build in public (even quietly). Writing about what you're building creates accountability.
Ship something every week, even if it's small. Momentum compounds.
Separate your identity from the product. Your app failing doesn't mean you failed. It means you learned something that cost you nothing but time.

The builders who win aren't the smartest. They're the ones who kept going when it stopped being fun.

What I'm Building Next

Rewrite Labs is still growing. The app suite is expanding. The OSS library count is going up.

If you're a student reading this wondering whether to start — start. The tools have never been cheaper, the knowledge has never been more accessible, and the market has never been more open to AI products.

You don't need permission. You just need to ship.

Follow along — I'll be posting technical deep-dives, build logs, and honest lessons from the trenches right here on Dev.to.

The Shadow Syllabus: How I stopped relying on my college degree (Case Study

IamAdhitya — Thu, 08 Jan 2026 15:45:26 +0000

The Double Life of an Engineering Student...

The syllabus ends at 5:00 PM. My real education starts at 8:00 PM.

I am currently a full-time engineering student.
I attend the lectures.
I respect the degree.
But I stopped expecting the university to teach me everything.

In my manual, The Rewrite, I analyze a case study named "Zahir" (Chapter 5).

He realized early on that a degree gives you safety, but self-education gives you leverage.

He didn't quit college.

He just added a second shift.
Day Shift: Thermodynamics, Assignments, Attendance. (The Script).
Night Shift: Psychology, Systems Engineering, Building Products. (The Rewrite).

I am living this "Double Life" right now. When I look around the classroom, I see many students waiting to be spoon-fed skills that will get them a job. But the market doesn't pay for what you were taught.

It pays for what you built while no one was watching.

If you feel like the curriculum is too slow for your ambition, don't blame the college. Build your own Shadow Syllabus.

Read Chapter 5 in "The Rewrite" (v1.0) here in our website - https://rewrite-labs.netlify.app/

The Syntax Trap: Why 2026 Belongs to Architects, Not Typists

IamAdhitya — Fri, 26 Dec 2025 09:54:11 +0000

Welcome to the first edition of Code & Context. Every week, I break down the intersection of Engineering, AI, and Human Psychology.

If you walk into any Computer Science lab in India right now, you will see the same thing: Students furiously memorizing syntax.

They are memorizing matplotlib libraries. They are stressing over missing semicolons in C++. They are treating coding like a vocabulary test.

As a CS student myself, I understand the pressure. But as someone who has spent the last year analyzing the AI landscape (and recently being recognized as a Top AI Creator for it), I have a hard truth to share:

In 2026, Syntax will be the cheapest commodity on the market.

We are entering a new era of software development. If your primary skill is "I know how to write a for loop without looking at documentation," you are in trouble.

Here is why the market is shifting, and what you need to do about it.

The "Copilot" Reality
Three years ago, writing a boilerplate API in Python took 20 minutes. Today, with tools like GitHub Copilot and Cursor, it takes 20 seconds.

AI has commoditized the "How." It can write the code, refactor the code, and even write the test cases for the code.

As the graph of AI capability goes up, the value of pure syntax memorization goes down. But notice the other line? That is Context.

The Rise of the "Architect"
If AI handles the syntax, what is left for us humans? The answer is Context.

The developers who will dominate in 2026 aren't the ones who type the fastest. They are the ones who can answer these three questions:

The Architecture Question: "AI can write the function, but where does this function fit in the microservices architecture?"
The Logic Question: "The model gave me three solutions. Which one is the most scalable, and which one introduces a security vulnerability?"
The Human Question: "How do I explain this technical constraint to a non-technical founder?"

This is why I named this newsletter Code & Context.

The "Code" is the easy part. The "Context" is the leverage.

Stop Being a Compiler
My advice to my fellow students and developers is simple: Don't stop coding. But stop learning like a compiler.

Don't just memorize the Pandas library. Learn data storytelling.
Don't just copy-paste the Neural Network code. Understand the math behind the weights and biases.
Don't just build the app. Document the decision-making process.

The future doesn't belong to the person who writes the most lines of code. It belongs to the person who knows why those lines were written in the first place.

What to Expect from This Newsletter
I am M. Adhitya. I operate at the intersection of hard engineering and human narrative.

In the coming weeks, Code & Context will cover:

Deep Dives: How LLMs actually "think" (without the jargon).
System Design: Building scalable architectures in the age of AI Agents.
Tech Strategy: How to build a personal brand as a developer.

If you are ready to move from being a Typist to being an Architect, you’re in the right place.

See you next week.