Founder-facing platform that aggregates financial data across accounting tools and spreadsheets into a unified view of current and projected cashflow, with deterministic, explainable insights.
- Sole frontend engineer
- Owned frontend architecture and SSR migration
- Led data modeling discussions for financial entities
- Designed and implemented dashboards and insights system
- Worked directly with founders on MVP scope and trade-offs
Early-stage founders often lack a clear understanding of their financial position because data is fragmented across:
- Accounting tools (QuickBooks, Xero)
- Spreadsheets
- Manual invoices and bills
Existing tools either focus purely on bookkeeping or rely on opaque AI outputs that are hard to trust for financial decision-making.
Shep separates financial data into two explicit layers:
- Current — actual, posted transactions that already happened
- Projected — invoices and bills representing expected future cash movement
From this, Shep generates deterministic insights such as:
- Overdue receivables
- Upcoming cash deficits
- Timing mismatches between income and expenses
The system avoids AI hallucinations and always produces the same output for the same input.
- Next.js App Router, SSR-first approach
- Convex backend for data ingestion, caching, and querying
- Integrations: QuickBooks, Xero, Google Sheets
- Deterministic, rule-based insights engine
The initial client-heavy MVP was rebuilt to use SSR, eliminating race conditions, inconsistent state, and data desynchronization issues that were impossible to debug in a purely client-rendered setup.
- MVP timeline and limited resources
- No payment processing (aggregation only)
- Multi-provider data normalization
- Financial correctness prioritized over “AI magic”
- MVP completed and actively evolving
- Core repository is private (startup-owned)
- Code access available on request