I'm finishing a BCA in Ganganagar, a city in Rajasthan that has never appeared on a tech company's map. I have no internship. No referral. No alumni network that ends in a warm intro. I have never completed a LeetCode streak.
What I have instead is an API running in production with paying-tier infrastructure, a sync protocol, and 116 tests.
I want to write about that choice honestly — including the parts that didn't work out — because most posts like this are written by people it worked out for, which makes them useless as advice.
The choice
The standard path was obvious enough. Grind DSA, do the puzzles, get good at the ninety-minute interview, hope a recruiter looks at a college they've never heard of.
I couldn't make myself believe in it. Not because it's wrong — it demonstrably works for a lot of people — but because I'd be competing on the exact axis where my disadvantages were largest. Every candidate from a tier-1 college with a referral was playing that game too, and they were playing it with better equipment.
So I picked the axis nobody was competing on: actually shipping things that other people could use.
Not clones. Not tutorials with my variable names. Things that run, that break at 3am, that have to be correct when nobody is watching.
What that actually looked like
It looked like being wrong a lot, in public, on a schedule nobody set.
I needed a list of exercises for an app I was building. There wasn't a decent one, so I built the catalog. Then the catalog needed to sync to a phone that loses signal in a basement gym, which turned out to be a genuinely hard distributed-systems problem hiding inside a fitness app. Then it needed API keys, so people couldn't just drain it. Then keys needed tiers. Then tiers needed billing.
Nobody assigned me any of that. Each thing was just the next thing that had to be true for the last thing to matter.
That's the entire method, and it isn't clever: build something you actually want to exist, then refuse to stop when it gets boring. The boring part is where the learning is. Anyone can build the fun 60%. The last 40% — errors, edge cases, the migration you can't roll back — is the part that teaches you what production means, and it's the part tutorials skip because it doesn't demo well.
What it cost me
Here's the part these posts usually leave out.
I can't pass your interview. If you sit me down and ask me to invert a binary tree in twenty minutes on a whiteboard, I will be slower than someone who has done four hundred of them. That's not a complaint about the format — it's just true. I optimised for a different thing, and the bill comes due in the room.
I have no team scars. I've never merged into a codebase with forty other people in it. Never had a design doc torn apart. Never inherited something awful from someone who left. Everything I know about working with others is theory, and I know it's theory.
Self-directed means self-inflicted. No senior looked at my first schema and said "that's going to hurt you in a week." I had to ship the mistake, feel it, and fix it. That's a real education, but it's an expensive one — the slowest possible way to learn a thing someone could have told me in one sentence.
Nobody is coming. There's no campus placement cell fixing this. If it works, it works because somebody read something I built and thought huh, this person can think. That's a thin thread, and some weeks it feels very thin.
What I think it bought me
I know what I don't know, precisely. I can tell you exactly where my API is
weak — the tombstone wart I can't fix without a breaking change, the webhooks
with no reconciliation job, the migrations with no down. That's not modesty.
It's the map. You only get a map like that from having to live somewhere.
I've made real decisions with real consequences. Not "which framework" — things like what does this API guarantee to a client that goes offline for a week. Get that wrong and someone's data disappears with no error message. Answering questions like that changes how you think in a way no exercise does.
I have something to point at. When someone asks what I can do, I don't describe it. I give them a URL and they run it. That's a fundamentally different conversation.
Would I recommend it?
Honestly? Only if you'd do it anyway.
If you're grinding DSA and it's working, keep going — this is not a moral argument and I'm not the guy to take advice from about a path I didn't take.
But if you're in a small city with no network, on a résumé nobody's going to shortlist, and the standard path feels like queueing for a door that isn't going to open for you — then the alternative is not a hack or a shortcut. It's just harder, and slower, and it works differently: you stop asking to be let in and start building something they have to look at.
I don't know yet if it worked. I'm graduating in August with a product, a blog, and no offer. Ask me in six months.
But I'd rather be judged on something I actually built than on how fast I can reverse a linked list — and if that costs me some interviews, I think I can live with that.