Supplier-Matching Workspace (R&D)
An SMB accounting/fintech platform
Overview
A working folder dedicated to the platform’s supplier-matching problem, linking OCR-extracted invoice vendors to canonical supplier records. The directory is effectively empty as captured here, so it is documented honestly as a scratch/R&D workspace that paired with the active company-matching notebook and the supplier-reconciliation logic that lives in the n8n workflows.
Why It Exists
Reliable supplier matching is central to automated bookkeeping: every ingested document must be attributed to the correct vendor despite noisy, inconsistent OCR names. This workspace was set aside for that effort.
What We Built
No retained artefacts in this folder. The substantive supplier-matching work is documented under the company-matching entry (fuzzy name matching with pandas and the name_matching library) and in the production Get business company_account and vat_code / Compare predictions with agent data n8n workflows. This entry is included for completeness and cross-linking.
Technologies & Approach
Intended for Python-based fuzzy matching and entity resolution, consistent with the sibling matching notebook.
Outcome / Impact
As a near-empty workspace it produced no standalone deliverable; the capability it represents was realised in the related matching notebook and the platform’s n8n reconciliation flows.
Capabilities Demonstrated
- Dedicated focus on supplier entity resolution within an accounting pipeline
- Honest tracking of exploratory workspaces alongside delivered work