Fyndel Construction Financial Intelligence Platform

Fyndel is a construction financial intelligence platform for companies that need a clearer view of WIP, budget movement, billing position, project risk, and forecast confidence.

The system combines financial workflows, secure tenant isolation, workbook intelligence, accounting integrations, and ML-driven forecasting so operational data can be turned into decision support instead of staying trapped in spreadsheets.

• Created a Next.js 16 and TypeScript application with protected routes, organization-scoped dashboards, entity detail flows, WIP reporting surfaces, customer/service-location workflows, team/workforce modules, and financial statement views.
• Modeled the backend around Supabase/PostgreSQL, Drizzle ORM, RLS, RBAC, audit trails, immutable financial snapshots, budget line items, integrations, and durable import tables.
• Designed the workbook import system as an AI-assisted pipeline that uses NLP, embeddings, entity resolution, clustering, and classification to extract projects, customers, invoices, cost codes, relationships, and review issues from multi-sheet Excel workbooks.
• Built Python and FastAPI compute services for ML inference, Monte Carlo simulation, feature generation, model evaluation, and tenant-scoped financial insight delivery.

The ML layer is built on top of tenant-scoped project, cost, billing, labor, workbook, and accounting data so the models can learn from the same operational records the product already uses.

Current model work includes cost-to-complete forecasting, margin erosion detection, project risk scoring, revenue leakage detection, over/under-billing analysis, and workbook intelligence for extracting structured financial entities and review issues.

Fyndel pulls together the strongest parts of my profile: product ownership, full-stack engineering, database security, financial modeling, applied ML, and the systems work needed to deliver model output inside a usable product.