Interviewee is powered by Gemini 3 Flash Preview, which serves as the core reasoning engine behind both interview preparation and live interview simulation. Gemini is first used to generate a personalized interview blueprint by analyzing a candidate’s resume alongside a real job description. From this input, Gemini produces a structured JSON output that identifies the key role competencies being evaluated, potential gaps between the candidate and the role, company-specific interview considerations, and the most likely interview format (behavioral, technical, or case). This reasoning layer is intentionally exposed in the UI, making the system transparent, explainable, and auditable for both users and judges. Gemini 3 is then used to orchestrate a live, adaptive interview experience. During the interview, Gemini dynamically generates interviewer questions and follow-ups based on the candidate’s responses, simulating how real interviewers probe, redirect, and assess answers in real time. After each response, Gemini evaluates performance using a STAR-based framework and produces a detailed scorecard covering clarity, structure, impact, and role fit. It also rewrites answers to demonstrate how candidates can improve. By leveraging Gemini 3’s fast reasoning, structured output, and low-latency generation, Interviewee moves beyond static interview prep to simulate real hiring judgment in an accessible, human-centered way.
Built With
- google-gemini3
- google-genai
- next.js
- tailwindcss
- typescript
- vercel
- webspeechapi
Log in or sign up for Devpost to join the conversation.