Public version of OCR pipeline used for agentic OCR, attaching to other systems or running as a standalone OCR service. The aim is to use GPU and CPU in combination. This should provide the best speed increase and allow for the maximum throughput. This repo is setup for large scale OCR processing at scale or local processing on single machines.
python kubernetes distributed-systems typescript ocr gpu helm named-entity-recognition celery ediscovery document-processing fastapi pii-detection legal-tech chain-of-custody paddleocr document-intelligence pdf-ocr electronic-discovery forensic-ocr
-
Updated
Jun 8, 2026 - Python