Case Studies

Our client is a digital transformation company focused on helping HR agencies automate their efforts with their workflow automation app, which includes handling of CVs, storing candidates' information and employer data, and more.

Looking to improve their system, the client approached us to create an AI-powered CV processing system which would automatically extract relevant information and assess candidate's skill level.

We have developed an AI module for processing digital CVs. The module extracts important data, like candidate name, desired compensation, skills, experience, etc. in a matter of seconds irrespective of document formatting.

ChatGPT For Entity Extraction

Entity extraction, or detection of specific data from a document, is performed by ChatGPT 3.5. We have assessed multiple LLMs on several criteria, like:

• availability, or how easy it is to get access to the model’s API and integrate into the project,
• processing speed, or how much time it takes the model to process one request,
• monthly cost, or how expensive the model is considering the average number of requests per month.

Based on the assessment, ChatGPT turned out to be the best choice for the task. The model detects and extracts entities form user requests and CV documents:

Query: I am looking for a help in C# development in Vienna for a data security project with 2 years of experience

Properties:

• Job Title: C# Developer
• Seniority: Junior+
• Industry: Data security
• Location: Vienna

After the user sends a query, the app extracts entities from candidate CVs, prescribes them a qualification level, like "junior", "middle", or "senior" levels for software engineer position candidates based on their skills and achievements.

Semantic Search

We have implemented semantic search to enable users to input vague search queries and get specific results.

After the entities are extracted, the app sorts through the relevant CVs and presents them to the user. We use Elasticsearch to sort through the candidate CVs and rank them according to how well they fit the search criteria.

With each new query, the search algorithm learns and improves, providing more accurate results over time.

The module sped up CV processing tenfold due to smart information extraction and smart candidate sorting by qualification, reducing the time it takes to find a candidate for any position, thus  making our client stand out among competition.

The module we have developed has been successfully implemented in our client's HR automation app, greatly increasing the analysis efficiency, reducing the CV processing time, and reducing the time it takes to interview each candidate due to smart skill assessment.