FinOps Brain is building the reconciliation cockpit for Indian CA firms. GSTR-2B vs Purchase Register matching is one of the most painful, high-stakes monthly workflows in Indian compliance — a missed match can cost a client their input tax credit. Our product ingests both files, runs a deterministic matching cascade, surfaces only the exceptions that need human judgment (with explanations and a full audit trail), and keeps the CA in control of review and sign-off.
The MVP is not a slide deck. It's built and production-deployed: multi-tenant, both file parsers, the deterministic reconciliation engine, audit logging, hundreds of tests, and CI/CD. We're now onboarding design partner CA firms and iterating on real client data.
You'd be joining at the exact moment the product meets the market.
THE ROLE
You'll be engineer #1, working directly with the founder (a Lead Data Scientist by background). You'll own large surfaces of the product end-to-end:
- Ingestion and parsing edge cases from messy real-world files
- Hardening the reconciliation engine
- Building the exception-review UI that CAs live in every day
- Shipping rapid against live design partner feedback
OUR STACK
- Backend: Python / FastAPI
- Frontend: Next.js, React, TypeScript, Tailwind
- Database: PostgreSQL (with pgvector)
- Async jobs: Redis + Dramatiq
- Infra: PaaS (Render / Fly.io) + Vercel, India region
- AI: Claude / OpenAI APIs as a rented intelligence layer on top of a deterministic-first core
WHO WE'RE LOOKING FOR
- A BUILDER FIRST. You've built something of your own — a side project with real users, a startup (even one that failed), or something you're hacking on right now. Failed founders are strongly encouraged to apply; that experience is an asset here, not a gap.
- IIT / NIT / BITS graduates preferred — but a strong shipping record matters more than any degree.
- 3+ years building and shipping production web applications, comfortable across the stack (ownership matters more than e