The CV Priority Sorter
Partner with us to enhance recruitment: test the CV Priority Sorter AI tool
Recruiting for positions with a high volume of applicants can be overwhelming and time-consuming. Sorting through countless resumes not only drains resources but also risks overlooking qualified candidates. On top of that, unconscious bias can unintentionally influence decisions, making it harder to ensure a fair and diverse recruitment process.
That’s why AcademicTransfer is developing the CV Priority Sorter—an AI tool that helps you efficiently rank anonymized resumes based on relevance. With this tool, you can save time by focusing on the most relevant candidates and reduce bias by assessing skills and qualifications without personal information.
To make this tool as effective as possible, we need your input. By participating in the testing process, you’ll help us fine-tune the CV Priority Sorter to better match the demands of real-world hiring, especially for job openings with a large number of applicants.
How does it work?
The CV Priority Sorter operates on a ‘swipe’ principle: a relevant resume is swiped to the ‘relevant’ side, while a non-relevant resume is swiped to the ‘not relevant’ side. After each selection, the tool reshuffles the digital stack of resumes, showing the most relevant one next. The further you go through the stack, the less relevant the resumes become.
The resumes are anonymized before you see them.
We started building this tool based on an existing program called ASReview, which does the same for scientific articles. Over the course of the project, we developed our own Large Language Model. However, the principle remains the same, and the video explaining the ranking process is still relevant.
What are the benefits?
- It avoids bias: The CV Priority Sorter tool anonymizes resumes, ensuring a fairer selection process by reducing unconscious bias. This allows employers to focus solely on candidates' skills and qualifications, promoting diversity and inclusion in recruitment.
- It saves time: Knowing that you will see fewer and fewer irrelevant resumes will help you scan them faster. The CV Priority Sorter also ensures that you don’t miss any relevant resumes in case relevant information might not be clearly presented, or because your focus wanes after reviewing many resumes.
How can you help?
We can only optimise the CV Priority Sorter by testing it. And for that test, we need to work closely with the hiring team because we need to understand the differences between your decisions and the model’s suggestions.
Recruiters can help by drawing attention to this tool and by facilitating the test set-up once a vacancy is included in a test.
Vacancy holders can help by involving us in a hiring procedure where many applicants can be expected (often PhD- or Postdoc-positions). The total time we need you to be involved will be 1.5 hours, in which:
- We explain the project, steps and expected outcomes.
- You select the relevant and irrelevant CVs - we require a list of at least 15-20 CVs marked as relevant and 15-20 as irrelevant from your perspective.
- We review the results with you and align them with the hiring team’s opinions.
What happens behind the screens?
All procedures that we run, will be run offline on a local system. We do not need to upload the CVs to a server. We require a memory stick, consisting of all the separate candidates’ CVs, which can be in PDF or Word format (there is no preference), but they must be in English.
The entire pipeline will run on our local system. Alternatively, we need a local system with Python installed and the relevant libraries (including ASReview). We’ll need to install these necessary libraries and set up the pipeline on the system, which could take about 30 minutes, excluding any experimental work.