← All AI buildsPRODUCT CASE STUDY · BUILT WITH AI

Artha360

A personal wealth-management platform for Indian households, with every investment, asset, loan, expense and goal in one place, plus a SaaS operator console to run it as a business.

Type
Own product, built with AI coding agents
Market
Personal finance, India
Clients
React web app; iOS and Android in progress
My role
Product, architecture, API design, review & testing
Python 3.13FastAPIMongoDBReact 19TypeScriptSwiftUIKotlin
API routes across 28 routers
184
backend tests passing
165/165
investment types supported
22
HIGHLIGHTS

Five things that set Artha360 apart.

  • Built for India

    PPF, EPF, NPS, SSY, SGB, FD/RD maths and Indian tax reports, alongside funds and stocks.

  • Whole family, one login

    Track every family member separately and filter any screen by person.

  • Live budgets

    Budget spend is calculated from real expenses, never typed in by hand.

  • Security built in

    Device and session control, login history, 2FA for admins and a full audit trail.

  • One API, five apps

    A locked OpenAPI contract serves web, admin, iOS and Android.

ARCHITECTURE OVERVIEW

One locked API contract behind every app.

Every client builds against the same OpenAPI schema. Customers and admins sign in on separate tracks, so an admin token never works on customer routes or the other way round.

ClientsBackendData & servicesWeb appiOS appAndroid appAdmin consoleREST · JWTFastAPI28 routersInvestments & fundsCash flow & budgetsReports & searchAdmin & billingMongoDBBeanie on async PyMongoOTP providermocked in developmentAudit logevery admin changeSeparate customer & admin auth184 routes · 165 testsAudit trail on every admin change
What makes it different

The same backend runs a consumer finance app and the operator console that manages its customers, plans and support.

WHAT THE PLATFORM DOES

A household’s whole financial picture.

  • Dashboard

    Eight KPIs, net-worth trend, asset allocation, top movers, upcoming payments and recent transactions.

  • Investments

    A 22-type instrument catalog with FD/RD maturity maths, plus mutual funds (XIRR, 1Y return) and stocks (P&L).

  • Assets & liabilities

    Purchase vs current value for assets; EMI, tenure, rate and repayment progress for loans.

  • Income & expenses

    A categorised cashbook with trend and category charts.

  • Recurring payments

    SIPs, EMIs, premiums, bills and subscriptions with a paid/upcoming record for every month.

  • Cash flow & goals

    Actuals with a forward projection, goals, and budgets tracked against real spend.

  • Reports

    Net worth, allocation, income vs expense, capital gains, insurance, loans, goal progress and tax.

  • Admin console

    Customers, advisors, plans, subscriptions, invoices, coupons, support tickets, staff roles and platform KPIs.

DESIGN DECISIONS

The options weighed, and the trade-offs accepted.

How the hardest calls were made: what else was on the table, why it lost, and what the chosen design costs.

Testing against a real database

The data layer moved from Motor to PyMongo’s native async driver, and mocking libraries lagged behind it.

  1. Mock the database driver
  2. Use an in-memory fake
  3. Spin up a real ephemeral MongoDB per test session (chosen)
Decision

Tests start a throwaway mongod and run every query for real.

Trade-off accepted

Slower test runs and MongoDB needed on CI, in exchange for tests that catch real query bugs.

TECHNOLOGIES & INTEGRATIONS

The stack behind it.

Backend
Python 3.13FastAPIBeaniePyMongo asyncArgon2JWT
Web
React 19TypeScriptVite
Mobile
SwiftUIKotlinJetpack Compose
Quality
pytestOpenAPI contract
Skills shown
Product designAPI designFintech domainSecurityAI-assisted development
FULL BUILD LOG

The work, piece by piece.

  1. Customer auth with signup, OTP verification, password reset, refresh tokens and logout, plus device and session management with forced logout.

  2. Permanent human-readable customer IDs (AR100001), per-platform device capture and a filterable login history.

  3. Investments ledger with a 22-item instrument catalog and FD/RD maturity maths; dedicated mutual-fund (XIRR, 1Y return) and stock (P&L, LTP) modules.

  4. Assets and liabilities with EMI, tenure, rate and repayment progress.

  5. Cashbook for categorised income and expenses, and recurring payments (SIPs, EMIs, premiums, bills) with a paid/upcoming log for every month.

  6. Cash flow with actuals and a forward projection; goals and budgets whose spend is computed live from real expenses.

  7. Dashboard with 8 KPIs, a net-worth time series, allocation, movers, upcoming payments and recent transactions.

  8. Eight report types: net worth, allocation, income vs expense, capital gains, insurance, loans, goal progress and tax.

  9. Multi-family-member tracking under one login, ⌘K federated search, and import/export.

  10. Super Admin console backend: customers, advisors, plans, subscriptions, invoices, coupons, support tickets, staff roles and a platform dashboard.

  11. Separate admin auth track with 2FA by default, rate-limited login, and audited, short-lived customer impersonation.

  12. Every admin change written to an audit log with who, what, old value, new value and IP.

  13. Tests run against a real ephemeral MongoDB instance rather than a mocked database.

Want a walkthrough?

The code is private, but I'm happy to demo the product and talk through the architecture. I'm available immediately for senior roles and contract work.