- Automated sourcing of candidate profiles and retrieving key information regarding their candidacies
- Enabled finding candidate profiles with desired specialization or qualification with 90% accuracy
- Enabled filtering of candidate profiles based on chosen criteria.
- Automated ranking of candidates from best to least suitable for particular positions ones
About the Project
ProvenBase — a US-based provider of B2B sourcing services and the owner of an intelligence sourcing solution. ProvenBase puts a great emphasis on Diversity, Equity, and Inclusion (DEI) practice and regards its principles as a core of the sourcing approach. Clients of ProvenBase are businesses from across multiple industries looking for ways to streamline the work of their HR departments and find diverse talents.
Challenges & Project Goals:
The customer was hoping to enhance the sourcing and candidate search capabilities of their intelligence product through candidate sourcing platform development, which should also enable them to move to the next investment round.
The goals of the project were to:
- Enable the business to find diverse talents with required expertise at low resource expenses and within reasonable deadlines
- Provide recruiters and HR specialists with advanced sourcing and search capabilities that would optimize time expenses and streamline hiring processes
- Help the company address diversity in workplace issues and meet their DEI goals easily
- Ensure seamless candidate sourcing and searching within the platform
- Make the solution user-friendly and semi-automated, helping recruiters to find the most suitable talents at low effort while at the same time build the solution in a way so it provides expanded search capabilities
We created an ML solution composed of several trained ML models, such as an AI-based job description analyzer and a sourcing algorithm, and integrated it into the existing IT system. We ensured the seamless work of the newly added modification with the digital environment that was built prior to the project’s start. We had been guiding every step of the project, from choosing the datasets for training ML models to selecting and integrating the features by the DEI principles. Besides, we were building the resulting functionality with the vision of our customer toward the ideal sourcing and candidate searching flow in mind.
Business Value Delivered:
Intelliarts team ensured that the intelligent sourcing solution has improved functionality. The latter makes it possible for ProvenBase to expand the reach of their platform and offer advanced sourcing and better candidate search capabilities to clients. With the modified intelligent sourcing solution, the company moved to the next investment round successfully. The work done by the Intelliarts team essentially helped ProvenBase to progress toward achieving its strategic business objectives.
The Intelliarts team ensures that the resulting AI recruiting sourcing platform has the following functionality:
- Automated candidate profile sourcing. The solution is capable of checking databases of job boards and social networks such as LinkedIn and retrieving candidate profiles from them. The system can provide key information regarding job seekers based on text descriptions in their profiles. Examples are years of experience, hard skills, soft skills, region of living, etc.
- Semantic search engine with search filters. The integrated search engine can help recruiters and HR specialists find candidates with the required qualifications or specialization by using text inputs. The technology also supports an array of search filters. Examples are common professional background, working experience, specific skills, region of living, and those that help to address the lack of diversity, e.g., particular gender or ethnicity, in the workplace.
- Position requirements and candidate profile matching feature. The solution can match the requirements that businesses have for particular positions with information in the profiles of candidates. Moreover, it can range job seekers from the most to least suitable ones based on how closely they meet the job requirements. This way, recruiters may always have a constantly updated selection of best-fit candidates without additional efforts needed.
Over the course of the recruiting sourcing platform development, the Intelliarts team faced a number of challenges. Among them is the absence of specific metrics for evaluating the performance of the trained model. So, our team was to investigate what textual information from candidate profiles should be retrieved for further matching with job requirements and propose a set of metrics to use.
The customer didn’t have trained models for matching. So, the Intelliarts engineers were to select algorithms and train models from the ground up, as there were no extensively-tested algorithms that would suit the project’s purposes. However, it only impacted the development time, as we managed not to compromise the quality.
Another worthwhile aspect is that it was difficult to obtain feedback from end users. Since the solution has a B2B nature and was in its early stages of development, there was a limited number of clients. So, developers couldn’t run a proper analysis of user feedback and progress toward the improvement of the solution this way. Still, extensive QA testing mitigated the issue.
Intelliarts team was able to essentially contribute to the project through the development and integration of the ML solution serving as an AI recruiting sourcing tool for semi-automated candidate profile sourcing and searching and the modification of the existing AI infrastructure. As a result,
- The search engine with an array of filters was built and integrated, and metrics for evaluating the performance of profile-matching functionality were selected.
- The solution shows over 90% accuracy for gender and ethnicity detection.
- The developed platform can now offer a set of diversity AI sourcing tools for recruiters. It also offers a complementary functionality that allows for quick, semi-automated job profile sorting and meeting a company’s DEI goals.
- The project’s goals were successfully fulfilled, which allowed ProvenBase to move to the next investment round.
- We made the recommendation to our customer to enhance the semantic search functionality in the next project’s stage by incorporating the capability to filter candidates based on specific experience criteria, such as narrowly focused projects or specific IT architecture.