Skip to main content
Danh Phan.
CLIENTDOCUMENT AI OCRWORKFLOW AUTOMATIONHuman Resources

Automated HR Employment Document Processing Pipeline

AI-assisted intake pipeline that extracts, validates, and archives candidate employment records and contracts with 98% accuracy.

Key Outcome: 97% faster onboarding intake
OCR pipeline dashboard showing document verification stages

Quick Facts

Timeline / Year
2026
My Role
Lead Engineer
Industry
Human Resources
Core Stack
Python, PaddleOCR, LangChain

The Business Problem

An HR department onboarding 200+ seasonal employees monthly spent over 40 hours per week manually transcribing citizen IDs, health certificates, and contracts into HRMS spreadsheets, suffering from frequent typographical errors.

Implemented Solution

Engineered an intelligent document intake pipeline with automated OCR preprocessing, LLM schema extraction, and human-in-the-loop review for low-confidence fields.

Role & Responsibilities

AI Automation Engineer: Implemented OCR image rectification, custom extraction prompts, webhook connectors, and validation rules.

Architecture & Implementation Details

Built an n8n processing pipeline that watches secure cloud intake folders, triggers OCR inference on GPU-enabled workers, normalizes Vietnamese accents and national ID numbers, and flags ambiguous records for human verification.

Verified Outcomes & Business Impact

Reduced candidate intake processing time from 25 minutes per applicant to 45 seconds, with an overall extraction accuracy exceeding 98.4% across 1,500+ processed documents.

Other projects sharing similar service domains or technology stacks.