AI Interviewer

Architecture

Inspiration

Hiring in tech has become increasingly inefficient. A single job posting can receive thousands of applications, forcing companies to spend tens of thousands of dollars on recruiting firms while still waiting months to find a qualified senior engineer.

At the same time, capable candidates are often filtered out by Applicant Tracking Systems (ATS) simply because their résumé formatting does not match automated parsing rules. This makes the process slow, expensive, and often unfair.

In an era where AI is accelerating many industries, we believe hiring should also become faster, more efficient, and more equitable for candidates.

What it does

We built an AI Interviewer Agent powered by the Gemini Live API.

The system conducts structured technical interviews with candidates in a professional and consistent manner. It dynamically generates interview questions based on the candidate’s résumé and the job description, runs a live code-sharing session, and evaluates the candidate using widely adopted technical assessment frameworks.

At the end of the interview, the system produces a structured evaluation and standardized score that employers can directly use in their hiring pipeline.

Most importantly, the system ensures that every candidate gets a real opportunity, regardless of resume formatting or ATS compatibility. The evaluation focuses on actual skills rather than document formatting or subjective interviewer impressions.

How we built it

The AI Interviewer Agent is built using the Gemini Live API, which enables real-time conversational interaction during the interview process.

The system analyzes the candidate’s résumé and job description to dynamically generate relevant interview questions. During the interview, the agent can conduct technical discussions, run a live coding session, and evaluate responses using multiple structured evaluation strategies.

After the interview is complete, the system generates a structured report and score summarizing the candidate’s technical performance and strengths.

Challenges we ran into

One of the main challenges was designing an interview process that feels natural and professional while still remaining structured and consistent.

Another challenge was generating relevant technical questions dynamically from both the résumé and the job description, while keeping the interview focused within a limited time window.

Ensuring that the evaluation remained fair and aligned with real technical assessment standards was also an important design consideration.

Accomplishments that we're proud of

We successfully built an AI system capable of conducting complete structured technical interviews, including discussion of experience, technical questions, and live coding.

The system produces a clear and standardized evaluation report, which can be directly used by companies as part of their hiring pipeline.

Most importantly, the platform removes common barriers such as ATS filtering and interviewer bias, giving candidates a fair opportunity to demonstrate their skills.

What we learned

Through building this project, we learned how powerful real-time AI systems can be when applied to structured workflows such as technical interviews.

We also learned that combining conversational AI with structured evaluation frameworks can produce much more consistent and scalable hiring processes.

This approach has the potential to significantly improve both candidate experience and hiring efficiency.

What's next for AI Interviewer

Next, we plan to expand the platform with:

deeper technical evaluation frameworks for different engineering roles
improved code-execution environments for live coding exercises
live system design interview
better analytics and insights for hiring teams

Our goal is to create a system where companies can run thousands of fair technical interviews in parallel, dramatically reducing hiring time and cost while giving every candidate a real opportunity to be evaluated on their skills.