AI-Powered CV Processing Solution for a Global Software Development Company

Challenge

Every candidate’s CV arrived in a different format. PDFs, Word documents, varied layouts. Recruiters manually extracted relevant information and rebuilt each CV in a branded template. A single CV took 3-4 hours. During pre-sales peaks, requests came in bundles of 5-10 CVs with tight deadlines.

Key challenges included:

• 3-4 hours of manual work per CV
• Inconsistent source formats requiring individual handling
• Urgent bulk requests with no scalable process
• Skilled recruiters stuck on formatting instead of evaluation

Main Goals

The project aimed to automate CV reformatting while maintaining client-ready quality.

The solution had to:

• Accept CVs in multiple formats (PDF, DOC, DOCX)
• Extract key sections: profile, skills, experience, education, certifications, languages
• Generate standardized Word documents matching the corporate template
• Deliver finished CVs via email to the requester
• Handle bulk requests without bottlenecks
• Provide a simple submission interface for non-technical users

Team Size and Composition

Edvantis assigned an AI specialist and an RPA developer to build the solution. The client’s recruitment team provided ongoing feedback and testing throughout development.

AI/Automation Specialist:

• designed the extraction pipeline,
• built n8n workflows,
• implemented Azure OpenAI integration for intelligent data extraction.

RPA Developer:

• developed the UiPath process for document generation, template mapping,
• automated email delivery.

The lean team combined AI and automation expertise to deliver a production-ready solution in two months.

Solution Delivered by Edvantis

Edvantis designed a modular AI extraction pipeline using n8n for orchestration, Azure OpenAI for intelligent data extraction, and UiPath for document generation.

Ingestion & Preprocessing

The system accepts CVs in multiple formats (PDF, DOC, DOCX) and automatically detects file type. Raw text is converted to Markdown to preserve document structure for AI processing.

Modular AI Extraction

Rather than relying on a single prompt, the solution processes six CV sections through specialized extraction workflows:

1. Profile summary
2. Technical skills
3. Work experience
4. Education
5. Certifications and courses
6. Languages

This architecture improves accuracy and makes each component independently tunable.

Output Generation

Extracted data is standardized into structured JSON. UiPath then populates a branded Word template using bookmark-based mapping and emails the finished document to the requester.

User Interface

A simple web form allows recruiters to submit requests with candidate details and upload files. The system sends status notifications at key processing stages, with error alerts for failed extractions.

Ongoing Optimization

Edvantis continues to refine the solution through prompt engineering, extraction logic optimization, and output validation, informed by direct feedback from the recruitment team.

We Achieved Great Results Together

The solution transformed a multi-hour manual task into a streamlined, mostly automated process.

Key Achievements:

• Reduced CV processing time from 3-4 hours to under 60 minutes per document 
• Enabled handling of bulk requests (up to 10 CVs) without delays 
• Delivered consistent, professionally formatted output every time 
• Freed recruiters to focus on candidate evaluation instead of document formatting 
• Provided hands-free execution with automatic email delivery 
• Established a feedback loop for continuous accuracy improvements