Products Decoded · Portfolio

Things I have built and shipped.

I do not just write product specs. I build the product, put it in front of real people, and keep it running. Here is the work, told straight: what each one solves, who it is for, and where it stands today.

Career tech · Consumer SaaS

Jobtune Live

The job search, run like a system instead of a guessing game.

jobtune.io
The problem

Looking for a job is a full time job with no manager, no process, and no feedback loop. People rewrite the same resume twenty times, lose track of where they applied, and never really know if a role is worth the effort before they have already spent an evening on it. The advice online is generic. Most tools stop at "here is a list of jobs."

What it does

Jobtune treats one job hunt as one operation. You point it at the roles you actually care about and it does the heavy lifting around each one.

  • Reads a role and reads you. It looks at a job and tells you how well you fit before you invest time, so you chase the right ones.
  • Tailors the resume to the role. Not a template swap. A version written for that specific job, ready to send.
  • Builds an apply kit. Everything you need to apply well in one place, so applying takes minutes instead of an evening.
  • Keeps the whole search in one tracker. What you targeted, what stage it is at, what is next. No more spreadsheet held together with hope.
  • Gives you a public profile. A clean, shareable page at your own handle that you can hand to a recruiter instead of a PDF.
How I thought about it

The core decision was to stop treating job hunting as a feed of listings and start treating it as a small portfolio of targets you manage. Once you have that spine, everything else hangs off it: ranking, tailoring, tracking, follow ups. The product gets out of your way and lets you spend your energy on the few roles that matter.

The other rule was honesty. The tool tells you when a role is a stretch. A career product that only ever says "go for it" is just noise, and people can feel that.

Built with
Next.jsMongoDBClaude (Anthropic)Clerk authRazorpay
Where it stands
Live in production
Real users running real searches
Full apply flow
Fit, tailor, kit, track in one place
Public profiles
Shareable page per user
12-month targetsProjected
10,000
Registered job seekers
5%
Free to paid conversion
₹1.5L
Monthly recurring revenue
30,000+
Apply kits generated

Goals, not results. They frame what success looks like at scale and the unit economics behind the model.

My role

Product, design, and build. I scoped it, designed the flows, wrote the code, shipped it, and I run it day to day.

Consumer AI · Culture and faith

Astrika Live

Classical Vedic astrology you can actually read, ask, and trust.

astrika.in
The problem

Vedic astrology is one of the oldest decision frameworks people still use, and most of the digital versions of it are bad. They are either vague horoscope filler with no real chart behind them, or technically correct but written in a language only a practitioner can read. If you actually want to understand your own chart, your options are a pricey one to one consult or a wall of jargon.

What it does
  • Builds your real chart. Astronomically accurate, the way a serious practitioner would expect, not a generic sun sign.
  • Explains it in plain language. What the chart is saying, written so a first timer gets it without losing what makes the tradition real.
  • Answers your actual questions. You can ask a specific question and get a grounded, chart based reading, not a fortune cookie.
  • Stays current. Readings account for where things stand today, not a frozen snapshot.
  • Respects the source. It works inside classical Vedic method rather than blending in whatever sounds nice.
How I thought about it

The whole product lives or dies on trust. An astrology app that quietly makes things up is worse than useless to the people who care about this, because they will catch it. So the bar I set was simple: the chart has to be correct first, and the explanation has to stay faithful to what the chart actually says. Get those two right and the writing can be warm and readable without becoming fiction.

The second call was format. A consult is a conversation, not a report. So Astrika lets you ask, not just read, which is what people were really looking for all along.

Built with
Next.jsMongoDBClaude (Anthropic)RazorpayClassical Vedic method
Where it stands
Live
In production with paying users
Product Hunt
Launched publicly
Ask, not just read
Question based readings
12-month targetsProjected
50,000
Charts decoded
8,000
Monthly active users
5%
Free to paid conversion
₹1L+
Monthly recurring revenue

Goals, not results. They reflect the organic, content led growth the product is built for.

My role

Founder and builder. Product, design, engineering, payments, launch, and support. End to end.

Consumer AI · Wellbeing

Saanjh Building now

A calmer place to talk through what you are carrying.

Plenty of people are carrying something they have nobody to say out loud to. Therapy is expensive, waitlisted, and still carries a stigma in a lot of homes. So most of it just stays inside.

Saanjh is an attempt at a softer first step. A quiet, private space to put words to what you are feeling and feel a little less alone with it, at any hour, without a form to fill or a judgment to brace for. It is not a replacement for care, and it will never pretend to be. It is the thing you reach for at 1am when there is no one else awake.

I am building it carefully and slowly, because this is the one category where moving fast and breaking things is the wrong instinct. More to share when it is ready.

First-year aspirationsProjected
25,000
People reached in year one
100,000+
Conversations held
40%
Come back within a week

Goals, not results. Success here is measured in people who felt heard and came back, not in revenue.

Selected work

Shipped at companies, and earlier ventures.

Not live apps you can open, but the work that built the judgment behind the products above. Stated at a level I can talk about openly.

Fintech · United States and Malaysia

MoneyLion

Insurance, payments, and an ad engine inside Copilot.

I built and lead a business unit across two teams in the US and Malaysia, for a platform of 23M+ users. The signature build is an industry-first advertising engine embedded inside Microsoft Copilot, where the conversation itself becomes a channel that drives clicks and revenue.

$9.6M ARR business unitRevenue per click $7 to $20Embedded in Microsoft Copilot

Logistics SaaS · B2B

Intugine · ePOD and Control Center

Predictive logistics in a single dashboard.

Built ePOD, an electronic proof of delivery product, and Control Center, a machine-learning platform that predicts delivery timelines, calculates rates, and tracks shipments in one place. Used by enterprise fleets, and backed by a bank for bill discounting for trucking companies.

Applied MLDelivery-time prediction15+ enterprise fleets

Payments · India

SabPaisa · Split Calculator

The end of manual settlement.

Built Split Calculator, an automated reconciliation product that replaced a fully manual settlement process. I also led the product rebuild through the RBI licensing process, so compliance was designed in rather than bolted on.

Automated reconciliationRBI licensingIndian payments

Product studio · Founder

CoolHax Labs

100+ products, from smart mirrors to IoT.

Founded and ran a product studio out of college, grew it to a 50+ person team, and shipped 100+ products for 100+ clients including government. Signature builds include Mirror 2.0, a computer-vision smart mirror on custom hardware, and a line of 3D smart charging lamps.

Mirror 2.0 · CV + custom hardware3D smart lamps · IoT50+ person team

Let's talk

If one of these is close to a problem you are working on, I would be glad to talk.

Forward-looking targets on this page are goals, not reported results.