AI-Powered Invoice Reconciliation System Developer

We’re looking for an experienced AI & Backend Developer to build a robust Invoice Reconciliation Automation System that extracts, matches, and verifies invoice data from vendor documents against internal purchase orders and payment records.

This project focuses on automating a high-volume accounting workflow for a retail enterprise by combining OCR, NLP, and backend automation to eliminate manual reconciliation errors and delays.

You’ll design an end-to-end pipeline that processes PDFs, scanned invoices, and emails, applies intelligent extraction and validation logic, and integrates with internal ERP systems for real-time reconciliation and discrepancy flagging.

Responsibilities:

Build a high-accuracy OCR pipeline using Tesseract, Google Cloud Vision, or similar.

Develop NLP logic to extract key fields like invoice number, total amount, line items, payment terms, and dates.

Implement fuzzy matching and rule-based validation to reconcile invoice entries with internal records.

Use Celery + Redis to manage document queues, retries, and background processing.

Build REST APIs using FastAPI for integration with ERP and dashboard systems.

Deploy and monitor the pipeline on scalable cloud infrastructure.

Implement confidence scoring and error handling to flag anomalies or mismatches.

Key Features:

Handles mixed input formats: emails, scanned PDFs, and bulk uploads

Invoice-line-level matching with purchase orders and delivery confirmations

Dashboard/API support for review and exception management

Supports multilingual invoice formats (English, Spanish, etc.)
Required Skills:

Python (3.9+)

OCR Tools (Tesseract, Google Cloud Vision API)

NLP: spaCy, Transformers (HuggingFace)

Backend: FastAPI, Redis, Celery

Database: PostgreSQL or MySQL

Experience with ERP integration (Odoo, SAP, NetSuite, or custom)

Familiarity with PDF parsing libraries (pdfplumber, PyMuPDF)