Founder Notes
Founder Notes · 17 November 2026 · 8 min read
Brand & Product

I Didn't Choose Astrology. It Chose Me

I am a recruiter, not a software engineer. This is what it took to build a Vedic astrology engine alone from Bangalore, what breaks at 3 a.m., and why I am still in the chair.

By Sandesh
8 min read
I Didn't Choose Astrology. It Chose Me
I Didn't Choose Astrology.
Astrology Chose Me.

It is 3 a.m. in Bangalore.
My phone lights up.

Someone I have never met is awake too, and something I built has just failed them.
A payment that didn't go through.
A link that leads nowhere.
A page that worked perfectly when I checked it an hour ago.

There is no support team to forward it to.
No engineer on call.
No one to say, "I'll look at it in the morning."

There is me, a cup of tea gone cold, and a laptop I had only just closed.

This is the part of building Supastro that never makes it into a launch post.
So here it is.

The résumé that makes no sense

For more than a decade I worked in Talent Acquisition.
Recruiting engineers. Interviewing thousands of professionals. Building teams.

Nothing about that screams, "Bro is going to build an AstroTech startup."

But hiring thousands of people teaches you something nobody warns you about.
People are patterns.

Fear becomes practicality.
Comfort becomes stability.
Self-doubt becomes logic.

After enough interviews you stop hearing the answers and start seeing what sits underneath them.
That is exactly what pulled me back to Jyotisha.

Not prediction.
Pattern recognition.

Growing up, I wanted to know why two equally talented people end up living completely different lives. That question never left.
So outside work, I studied. Not Instagram astrology. Not "Mercury is ruining your life." Parashari principles, Jaimini techniques, nakshatras, dashas, vargas, yogas.

I read charts for friends.
Then friends of friends.
Then complete strangers.
One became five. Five became fifty.

And the reaction was never "Wow, that's magical."
It was, "How did you know that?"

The question I couldn't put down

Then I looked at the apps.

Enter your birthday.
Get your Sun sign.
Read a paragraph that somehow also applies to eight million other people.

That is not Jyotisha.

Real Jyotisha is mathematics. An exact birth time, a place, the sky at that minute, and then layer upon layer of computation on top.

Why wasn't anyone building it properly?

I carried that question around until it stopped being a question and became a decision.
I would build it myself.

One small problem.
I am not a software engineer.

I thought I was building an app

I was building an observatory, a library and a courtroom.

The observatory. Every chart starts from where the planets really were at the minute you were born, worked out from NASA JPL ephemeris data in the sidereal zodiac, on the Lahiri ayanamsa. Not from precomputed tables. A few minutes of error in a birth time can move a rising sign, so I don't get to be casual about the sky.

The layers. Not one chart. The birth chart and sixteen divisional charts, with more beyond them. Planetary strength measured from six separate sources. An Ashtakavarga of 337 points. Dasha periods computed four levels deep. All seven Jaimini karakas. Thousands of numbers for one person, before a single sentence is written.

The library. The classical texts, Brihat Parashara Hora Shastra, Phaladeepika, Uttara Kalamrita and more, broken down verse by verse, so that when the engine applies a rule it can point to the line it came from.

The courtroom. This is the part I am proudest of. The maths decides first. The AI is only allowed to explain what the maths found. For the big questions, three specialist agents argue over your chart, an independent verifier checks them, and a chief has to deliver the answer with any disagreement left on the record. And before an answer reaches you, a separate checker holds its numbers, dates and placements against your computed chart. What the chart cannot support gets rewritten or removed.

I learnt every piece of that the slow way.
Databases. Sign-in. Payments in two currencies. Deployments. Testing.
Not from a course.
From breaking it.

Everything works. Then it doesn't.

Here is the cycle nobody prepares you for.

You work all day on one thing.
It works.
You test it. It works again.
You release it. Green checkmark.
You make tea.

An hour later it is broken.
Not the thing you changed. Something three rooms away that you never knew was leaning on it.

I keep a small, strict collection of these.

The database that froze. More than once, and usually at night. One night my own fix made it worse: it kept restarting the thing before it had finished waking up. So I taught the system to check its own pulse every few minutes, to wait when waiting is right, to restart itself when it isn't, and to send me one email when it is over. The subject line reads: Supastro is back. I have never been so happy to get an email from myself.

The dasha that was "wrong". A member wrote to say their periods didn't match what they had seen elsewhere. I went through the maths for hours. The maths was right. The birth time saved with us said afternoon. The one they had used everywhere else said morning. One letter, A or P, had moved the rising sign and shifted the whole timeline by years. The engine wasn't wrong, and it was still my fault, because nothing on the screen had shown them the time they'd entered. Now it sits at the top of every screen in the dashboard.

The woman who knew better than my clock. She was born in Ohio in the 1960s and told me her chart was off. I was sure she was mistaken. She was right. In those years American towns chose their own daylight saving, and the standard clock records only remember what the big city did. Her chart had been cast an hour early. So had the chart from every other calculator I checked. I went through old birth records, corrected the clock for her part of Ohio, and then ran fifty thousand random births through the engine to prove I had not moved anyone else's chart.

One person. One hour.
A chart is somebody's life on a page. "Close enough" is a wrong chart.

So today nothing goes live until more than a hundred sets of tests pass. Most of them exist because something broke once.

One chair

People ask how big the team is.

The team is one chair.

Code. Testing. Deployments. Sales. Marketing. Branding. Social media.
Support at midnight, at 1 a.m., and again at 3.

The person who replies to your message about a failed payment is the person who wrote the payment code, and who will be up fixing it after he replies.

Some days it is heartbreaking, and I'll say that plainly.
You give something hours and watch it break in a second.
You fix it, and something else falls over.
There are letdowns I won't list here.

Some days I wonder why I voluntarily chose one of the hardest intersections imaginable:
AI.
Astronomy.
Ancient Sanskrit texts.
Startup life.
And users who expect everything to load in under two seconds.

Why I am still in the chair

Then a message arrives that isn't a complaint.

Someone saying the platform helped them understand themselves in a way nothing else had.

And the frozen database, the broken link and the 3 a.m. phone all go quiet for a minute.

AI became my accelerator, not my replacement for understanding.
It helped me move faster.
It never removed the responsibility of learning how everything fits together.
I still had to learn it one broken thing at a time.

I'm building this from Bangalore.
Alone.
Mostly with curiosity.
Mostly with stubbornness.
And definitely with far too much tea.

Not fortune telling.
Not fear marketing.
Not "buy this gemstone or bad things will happen."

Real computation, standing on the classical texts.

If you've ever rolled your eyes at a vague horoscope...
If you believe an ancient system deserves modern engineering...
Or if you've ever had an obsession that made absolutely no sense on your résumé...

You're probably my kind of person.
Come say hi.

It will break again tomorrow.
I'll fix it again tomorrow.

Because sometimes your biggest career move isn't climbing the ladder.
It's finally admitting what has been quietly calling you all along.

And at 3 a.m., with cold tea and a red error on the screen, I still wouldn't have it any other way.