{"message":"PDF OCR Confidence API - Performance Optimized","version":"4.3.0","features":["Mixed source processing: S3 files, uploads, or both simultaneously","Maximum performance PDF processing (1.5x DPI, 85% JPEG quality)","Parallel S3 downloads with concurrent upload processing","Memory-aware streaming processing for large PDFs","Intelligent batch sizing based on available memory","Dynamic batch sizing (up to 40 pages) for maximum throughput","Intelligent mode switching (batch vs streaming)","Dynamic concurrency based on system resources","Intelligent response caching with MD5 hashing","Connection pooling for AWS clients","Adaptive timeout management","Real-time performance metrics with detailed timing","Production-grade error handling","Background job processing with polling support"],"endpoints":{"/api/ai-ocr-checks-service/analyze":"POST - Process PDFs (supports async_mode parameter)","/api/ai-ocr-checks-service/status/{job_id}":"GET - Get job status for polling","/api/ai-ocr-checks-service/jobs/stats":"GET - Get job statistics","/health":"GET - Health check","/metrics":"GET - Performance metrics and system stats","/frontend":"GET - Frontend UI for PDF processing","/docs":"GET - API documentation"},"performance_optimizations":{"mixed_source_processing":"Parallel S3 downloads and upload processing","intelligent_processing":"Automatic batch vs streaming mode selection","dynamic_batch_sizing":"Up to 5 pages per batch (75% improvement)","memory_aware_processing":"Adaptive processing based on memory pressure","dynamic_concurrency":"Auto-scaled to 20 general, 10 Textract","response_caching":"MD5-based intelligent caching for cost efficiency","connection_pooling":"Reusable AWS client connections (S3 + Textract)","adaptive_timeouts":"Dynamic timeouts based on batch size","streaming_support":"Memory-efficient processing for large documents","background_processing":"Long-running jobs with polling support"}}