Avrio is a recruitment platform that facilitates the hiring process for recruiters enabling them to select the best talent that fits their requirements. It streamlines the whole process including candidate sourcing, screening, and interview scheduling.
Avrio enhances the recruitment process for both candidates and hiring teams. For candidates, it delivers a personalized unbiased experience instead of traditional lengthy application forms that might leave the candidate without a feedback where he feels disrespected and ignored. For the hiring teams, it helps them find top talents quickly and efficiently instead of going through ineffective interview process and getting poor quality hires.
Challenges & Opportunity
Avrio noticed the gaps and shortcomings with traditional recruitment processes, from increased time-to-hiring, to low quality hires and inefficient interviews. They wanted to deliver a solution that creates an automated and effective hiring pipeline for both candidates and recruiters.
Using a proprietary software algorithm, they built a platform that can match global talent with job opportunities. The platform analyzes the job requirements and different candidate skills, then it provides a "fit score" that promotes candidates who have higher matching skill set for the specific job requirements, ensuring the highest quality talent is accessible for the hiring teams.
The platform also delivers a personalized experience for candidates including a "candidate gap analysis" that shows which of their skills fit the job and where their gaps exist, solving a common problem with traditional non-transparent hiring processes.
However, recruiters and hiring managers on the platform wanted more tailored candidate results with even higher accuracy with the matching criteria, they also wanted to reduce the time it takes for the platform to analyze the candidates and report the result with the highest fit score.
Additionally, the complexity of the deployment and maintenance for the large software codebase was a challenge for Avrio, especially that it relied on a traditional infrastructure based on Elastic Beanstalk.
At the same time, they've been aware of the rising use cases and advancements in LLMs. So, they perceived this as an opportunity to modernize their platform and revolutionize the digital recruiting by utilizing LLMs.
Goals
Avrio wanted to boost the performance and accuracy of the platform for handling a massive scale of candidate-to-job profile matchings by refactoring their core solution to enable access to AI models for on-demand analysis of the job and candidate details, all while keeping operational overhead, deployment complexity, and cost at minimum.
Solution
Avrio partnered with Aland Cloud to achieve these goals and enhance their platform using AWS technologies. We were able to modernize the solution by integrating Amazon Bedrock AI as well as migrating parts of the architecture from Beanstalk to ECS. Here's a breakdown of the complete flow and the different components of the system after the modernization we implemented:
- A recruiter submits a job to the platform with all the details like job title, required skills, preferred skills, education, and other parameters that will be used for matching candidates. This submission can be done manually where he fills these different fields with the appropriate data, or he can submit a PDF document that already includes all these details.
- The web application, which is now hosted on ECS, stores the job details on S3 as a PDF document. If the job information is provided manually by filling the fields, a PDF document is generated from this information first. If the job was already provided through PDF, it is stored directly in S3.
- When the file is uploaded to S3, an S3 event triggers a lambda function which pulls the file and uses Bedrock to analyze it and parse the job metadata.
- The job metadata extracted from the PDF is used for comparison against candidate CVs and information. Some of these candidate information is stored internally, while others are retrieved from external APIs like Nexxt. We've also improved the query to Nexxt by including more details for a higher accuracy of retrieved candidate information.
- The filtered candidates' CVs and information is stored in DynamoDB against the job description.
- From this data insertion, DynamoDB streams is used then to trigger another lambda function. This lambda function again uses Bedrock to compare the CVs against the job description but with additional information and dimensions like candidate current and previous job positions, experience, additional skills, etc.
- After each comparison, a fit score is generated for each candidate against the specific job and stored in another DynamoDB table.
- Finally, the recruiter gets a list of highest matching candidates with their fit score. He can download their CVs, schedule interviews, and invite them for further hiring process steps.
Outcomes & Results
The platform was able to deliver better performance with higher accuracy of the results after this modernization. The deployment and maintenance tasks are also dramatically simplified. Avrio clients now are reporting:
- 40% increase in hiring productivity
- More accurate fit-score that matches higher quality talents
- Faster candidate results which are displayed within ~10 seconds compared to ~4 minutes reported previously
Conclusion
We were able to modernize Avrio's recruitment platform by leveraging AI capabilities from Amazon Bedrock, in addition to migrating some of the components from traditional infrastructure on Elastic Beanstalk to ECS. This modernization satisfied the client requirements for providing improved platform performance with higher accuracy of candidate-to-job matching results.
Recruiters now report more hiring productivity with the ability to find top talents globally without the need for traditional long hiring processes and ineffective interviews. They were also able to automate different administrative tasks like job screening and interview scheduling, all while delivering a personalized and positive experience for candidates that includes effective communication and feedback.
