Inspiration
Currently the hiring process involves a lot of man-power which is very expensive to organizations. The process begins with initial phone screening round done by HR where in the recruiter HR tries to understand the candidate's potential with reference to the job opening and required skill-set and makes a decision , if the candidate is fit for the job and sends him/her to the next round of interviews.At times even after going to the next round candidates are rejected or even after getting the offer they decline and the recruiter continuous with the search for another candidate for the job. The total cost involved and time involved is very high. Companies need to hire several recruiters to meet the increasing job market demand ,which involves a high cost especially in the form of salaries and training.
The tool AI-Recruiter is a solution to this problem. The tool uses Gemini-AI technology where in an AI-bot would be used for phone-screening of candidates. The tool provides a low-cost solution to organization for the hiring process in an complete automated way.
The AI based tool removes the need to human intervention in the initial screening process in recruitment with the aim of saving cost and time to organization and make quicker and more accurate decisions. As the screening process is automated it would help organization focus more on other areas besides recruitment.The AI tool would talk to the candidate similar to a human talking on phone-screening round thereby the interview experience for the candidate will not change and he would have the same experience as to interacting with a human on call.
1 human recruiter would interview 4-5 candidates in a day on phone with each interview being around 45-60 mins long but with the AI-tool multiple hundreds of interview could take place simultaneously just in 1 hr.
What it does
The tool AI-Recruitment is a python based tool that uses Gemini-AI Api. The tool needs the job description and candidate's name as input. The tool would get response from Gemini and talk to the candidate, similar to a human talking for the initial phone screening round.Based on the job description the tool would ask some basic screening questions to the candidate and record the candidate's response. Based on the response and the job description the tool would summarize the interview and make a decision if the candidate should be selected for the upcoming rounds.
How we built it
The code base used in python with the Google gen-AI tool Gemini. This being the prototype of tool the main focus is to develop a small functionality describing the phone screening round of hiring process. Algorithm
- Input the job description and candidate's name in the system
- The tool will make an API call to gemini with respect to the job description and read a short job summary to the candidate.
- The tool will then again make an call to gemini api to come up with questions to be asked to the candidate.
- The tool will now choose 3 questions from the list (as this is a prototype) and ask the questions to candidate and the candidate will ans them by speaking, the response would be recorded.
- Once the candidate has answered the question the tool would select the next question and would ask the candidate and response would be recorded and this would continue.
- The tool will send the recorded response and the job description to gemini to decide on the candidate's potential and respond back with a decision.
Challenges we ran into
Gemini would only return data in json format, the challenge was to parse it and then use the sound recognizer of python to say the response ,similar to a human eliminating extra words.
Accomplishments that we're proud of
Designing using Gemini API and incorporating it in python and learning the use and effects of GEN-AI The system working and providing result set with description as to why a candidate should be selected or not.
What we learned
Learnt to use google api with python learn basics of python. Learnt basics of AI and understood the working behind models and creating models and training datasets. Exposures to CNN .
What's next for AI-Recruitment
The current code base is prototype showcasing an idea with a concept. The future scope would include expanding this idea the concept would remain the same where screening interview sessions are done using AI. In future scope the questions asked by the AI model would come through a defined model specific to interviews which aligns more to the job role and some questions would be created based on candidate's resume. The AI model would be smart enough to counter-question the candidate as well as negotiate salaries and any other condition. The main aim of submitting this idea is to show case to use of AI in the corporate industry for recruitment and on enhancing this system we can get a full fledged working AI-recruiter with more working potential and a big cost saving for the organization.
Log in or sign up for Devpost to join the conversation.