VU
Resume
Back to Projects Showcase
OCR & Document IntelligenceProduction Deployed

KWC OCR + PDF Splitter

High-Throughput Autonomous Document Segmentation — Automated optical segmentation pipeline that identifies patient boundaries in multi-page clinical records, extracts demographics, and routes verified encounters to destination SFTP servers.

PythonTesseract OCROCR PipelinesPDF ProcessingSFTP AutomationKeyword-Based Document Splitting
Batch Ingestion & Split
< 6 min Batch
PDF Batch
370+ pgs
150 Pts
98.5% Cut
PT_001
✓
PT_002
✓
PT_003
✓
Manual: 4–6 hrsSaved: 2–3 min/patient
Measured Production Impact
98.5%
Time Reduced
4-6h down to <6m
< 6 min
Batch Speed
vs 4–6 hrs manual
2–3 min
Saved / Patient
Zero manual touches
100–150
Patients / Batch
200–370+ pages

The Problem

Clinical operations received massive consolidated PDF batches containing 200 to 370+ pages representing 100–150 distinct patients. Medical scribes and billing staff spent 4 to 6+ hours every day manually reviewing boundaries, cutting PDFs, renaming files, and uploading them to disparate SFTP directories.

The Automated Pipeline

Engineered an intelligent OCR-driven segmentation pipeline that parses multi-page medical records, detects individual patient header boundaries, extracts demographic anchors, and programmatically splits the master document into verified patient encounters.

Results & Reliability

Reduced batch execution from 4–6 hours to under 6 minutes with 98.5% turnaround reduction. The system operates fully touchless, directly transferring structured patient packets to downstream SFTP repositories with zero human intervention.

Key Engineering & Operational Highlights

Turnaround Reduction (4–6+ Hours → <6 Mins): Slashed processing time for high-volume batches from 4–6+ hours of manual document splitting down to under 6 minutes of automated background execution.
2–3 Minutes Saved per Patient: Eliminates manual boundary identification, multi-page verification, file extraction, renaming, and SFTP folder routing.
High-Volume Batch Capacity: Ingests consolidated PDFs averaging 200–370+ total pages (~2 pages per patient) and autonomously segments each encounter.
Zero-Touch Ingestion & SFTP Routing: Organizes patient-specific files into required folder structures and routes them automatically to destination servers.
← All ProjectsNext: Code Reconciliation System