Show HN: I built a forensic accounting engine for divorce My friend was quoted $50k by a forensic accountant to trace "separate property" (inheritance) through a joint bank account during a divorce. I realized the math they use (Lowest Intermediate Balance Rule, or LIBR) is deterministic, so I automated it. Bank statements are messy (scanned, skewed), and proving "Chain of Custody" is critical for legal admissibility. A Django app that ingests PDFs or shots and uses Mistral OCR-3 to parse transactions (way better reliability than Tesseract for tables), and runs the LIBR algorithm to identify exactly which dollars belong to whom. Recent Updates (Based on HN Feedback -- https://news.ycombinator.com/item?id=46350044): I just refactored the backend to support a strict Chain of Custody: Immutable Reports: Reports are now snapshot-locked upon generation. Cryptographic Integrity: Every report generates a SHA-256 hash of the raw calculation vector + the final PDF. This means you can mathematically prove the report hasn't been tampered with since generation. LIBR Verification: Added a regression suite to handle edge cases like "zero-balance dips," ensuring the tracing algorithm adheres to the See v. See (1966) legal standard. The code isn't fully open-source yet, but it's free to use right now. I'd love feedback on the PDF generation logic or the OCR accuracy! |