Interview Automation w/ Generative Ai

This is a end-end working automation that practically applies Gen Ai to the usually manual tedious interview & evaluation process.

BankNext’s rapid expansion requires embarking on major talent hiring exercise. Unfortunately, a huge volume of interviewed applicants turn out to be below par & hence rejected. This causes enormous time & monetary wastage. BankNext wants to change that. It needs a solution to rapidly pre-screen applicants and consistently shortlist only the right talent.

Pre-screening interview

A pre-screening interview strives to gauge, candidate’s overall competency level. The evaluation is designed to check if the candidate can pass at least this minimum acceptable level. Generally comprises of standard questions w/ deterministic answers. Only those applicants who pass this level should be shortlisted for the subsequent detailed interviews.

Current problem & challenges

1. Prescreening process is manual.
2. Requires senior engineer to dedicate min 45 mins/day to interview.
3. Candidate evaluation process is manual.
4. The success rate in selecting the right fit candidate is painfully low.
5. BankNext is falling behind due to lack of talented workforce.
6. Pre-screening candidates by 3rd party vendors has proved unreliable.
7. MCQ tests like Mettl are found to be quiet restrictive and expensive.
8. Houston, we have a problem!

Solution : Automated-Interviewer-Live-Candidate-Prescreening w/ GPT

1. Automation launches a text to speech engine.
2. Conducts interviews w/ a live candidate in real time.
3. Open AI’s GPT is employed to dynamically evaluate the responses.
4. Application talks to the candidate in a natural spoken language.
5. Candidate’s spoken responses are captured for evaluation.
6. MS Teams/MS Word generates the full interview text transcript.
7. Transcript + context + guidance data + selection criteria is fed to GPT.
8. GPT responds w/ detailed explanation after select/reject this candidate.
9. Thus the entire pre-selection completes w/o any manual interviewer.

System Flow Diagram

GitHub

1. Repository link & ReadMe [ OpenAi key + text2speech instructions ]
2. Sample interview text transcript
3. Sample candidate evaluation

Demo Video

Interview Automation w/ Generative Ai — YouTube

Conclusion : Automation Capabilities

1- Solution diagram depicts the entire interview automation flow.
2- Interview profiles can be dynamically switched via configuration.
3- Questions picked at runtime based on the profile configured.
4- Switch btwn Real/Mock interviews by simply flipping the config.
5- Interview duration adjusted via config eg. limitNumQnsToAsk=7
6- New interview profiles can be added w/o any code change.
7- Create a new profile dir w/ the question set.
8- Set the config param. eg. profile=software_engineer
9- Dynamic evaluation & ranking w/ Open AI’s GPT 3.5-turbo invocation.
10- Interviewer fully controls the selection criteria.
11- Specifies textual instructions to GPT at runtime. No code change.
12- Saves BankNext a ton of time, money & frustration.