About this prototype
Project 4 of the Mridula Initiative — a personal effort to deploy patient capital toward livelihoods around Kanpur, Uttar Pradesh.
I haven't yet started deploying real capital across SHGs, so this app ships with a fictional demo dataset and stores any edits in your browser only. The goal is to feel the shape of a real portfolio tool — the dashboard, the drill-down, the repayment ledger, the cashflow projection — and iron out the UX before any rupees are at stake.
What's here
Total disbursed, outstanding, on-time repayment, PAR-30/60/90, money multiple, XIRR, principal-weighted days outstanding.
Loan history, goals, and check-in notes for each Self-Help Group. CRUD on SHGs and on the loans inside them.
Mark scheduled installments paid, unmark them, or restructure via amendments. Partial payments and write-offs are first-class.
Slide default rate, interest haircut, and reinvestment toggles to stress-test the current portfolio out 12–60 months.
The data is fake
All SHG names, member names, contact persons, and amounts in the seed dataset are fictional. The villages and blocks (Akbarpur, Bilhaur, Ghatampur, Sarwankhera) are real Kanpur Dehat geography but the SHGs themselves are not real groups. Anything you edit is saved to your browser's localStorage only — refresh and your changes are still there; open the URL in another browser and you'll see the original baseline.
What's next
When I start deploying real capital, I'll build a v2 of this app with Supabase Auth + Row-Level Security so my real portfolio data stays private. The data shape in this prototype is the spec; v2 mostly adds an owner_id column to each table and the policies that go with it. Document upload, multi-user sharing with NGO partners, and a WhatsApp share path for monthly statements are also on the v2 list.
I wrote up the framing flip from “public sandbox” to “private tracker”, the localStorage-first design, and the data-modelling decisions on the blog:
Weekend 3: A Micro-Loan Portfolio Simulator