Web dashboard
LIVE
PelicanX
A weekly India fintech dashboard — payments, lending and wealthtech in one place, every figure marked with where it came from.
My Projects — 02
Things I built, and what each one taught me
Open the live version, or read the story behind it — problem, what I shipped, what I'd do differently.
Mobile web app
LIVE
GharOS
A household-staff ledger for Indian homes: who came, who's owed what, and when they're off — without the argument at the end of the month.
Web game
LIVE
Founder's Seat
A grand-strategy simulation for future founders: run a post-seed SaaS start-up for 36 months from one command dashboard, and get graded on your judgment, not your luck.
The problem
India's fintech numbers live in a dozen places — NPCI's monthly release, PIB notes, RBI circulars, company disclosures — and get quoted third-hand with no date and no source attached. If you want to know what UPI volume actually did last month, you either trust a headline or lose an afternoon.
What I built
- A weekly dashboard split into the three parts of the market that behave differently: payments, lending & credit, and wealthtech.
- A quick-numbers row per section — UPI volume and value, the FY total, P2M share, banks live, app market share, and the deadline the sector is actually watching.
- Charts you can superimpose: put All UPI, PhonePe, Google Pay and the rest on the same axis over a quarter, a year or five, and hover for the monthly value.
- A verification mark on every figure — ✓ where it is source-verified, ≈ where it is the latest reported estimate. Where a series is interpolated, the note under the chart says so.
What I'd change
Compilation is still manual, which caps both how often it refreshes and how far back the history goes. The next version should read the NPCI and PIB releases itself, so the weekly update stops being somebody's evening.
The problem
Most Indian households run on an informal arrangement with the people who work in them — a cook, a maid, a driver — and no record of any of it. Attendance is remembered rather than written down. Salary gets worked out at the end of the month from memory, minus some number of days nobody agreed to. That is the same argument, every month, in millions of homes.
What I built
- A shared record of days worked — marked by the employer or checked in by the helper with a QR code — so the month ends with a fact instead of two recollections.
- Work and rate written down per job: cooking, mopping, washing, the school run, driving, each with its own amount and time. Salary is the sum; a deduction is visible and optional rather than assumed.
- Leave requests, messages and the monthly payment record in the same place, plus a matching flow for finding help nearby.
- Everything stays in the browser on that device. No account, no server, nothing uploaded — which, for something handling ID and salary, is the design decision rather than a technical footnote.
What I'd change
ID capture is simulated and the matching side has no supply behind it — both are honest prototypes of a flow, not the thing itself. The next step is one real household running it for a full month and telling me what breaks.
The problem
Case studies teach start-up decisions in hindsight: you read what the founder did and argue about it. What they rarely teach is the feel of the job — that every choice spends the same scarce cash and attention, that hires and fundraises take months to pay off, and that you have to commit before the numbers are clear. Most MBA students and founder's office hires meet that for the first time on the job.
What I built
- A Europa Universalis-style command dashboard for a post-seed logistics SaaS company in Bengaluru: ten panels for finance, growth, product, team, operations, partners, capital, reports and goals, all driven by one simulation.
- A monthly loop where founder focus is the binding constraint: Strategy, Ops and People points refill every month, so you can raise, recruit or ship — rarely all three.
- Real operating texture: marketing channels with saturation and ROAS, A/B tests with noisy results, partner negotiations, enterprise RFPs that pull engineers off the roadmap, crises, board meetings, and a month-end P&L, cash flow and MIS pack.
- A debrief that grades every decision on what you knew at the time, so a good call that went badly still scores well. Shared seeds let a whole class play the same market.
- Accounts with Google sign-in, a 7-day free trial and a one-time unlock through PhonePe, plus an owner dashboard for sign-ups, daily logins and payments.
What I'd change
The balance is tuned with scripted bots, not real players, so the first cohorts will find the exploits. Next is a faculty view of a whole cohort's decisions, and a shared market where classmates compete for the same customers.