2026 – Present · In development
E-Statement Normalizer
A Python foundation for normalizing financial records with exact-decimal values, provenance, and typed validation.
- Context
- Independent project in development
- My role
- Data modeling & validation
- Tools
- Python · Decimal · JSON Schema
The problem
Bank and wallet statements differ in structure. A useful normalization layer needs to preserve the meaning of financial values and make it possible to trace a normalized record back to its evidence.
Implemented foundation
I implemented canonical financial data models, exact-decimal money handling, stable identifiers, provenance and evidence models, JSON Schema, and typed validation.
Synthetic tests exercise the data contracts and decimal round trips. These checks support the foundation; they do not establish real-statement extraction accuracy.
Design decisions
Unknown layouts and validation failures should remain visible for review rather than silently entering downstream records. Provenance and explicit validation make it easier to inspect how data reached its normalized form.
Status
Real-document ingestion remains in development. The project is not presented as a finished PDF ingestion service or a production ledger integration.