Mock Ai Interviewer

We built AI Mock Interviewer because text-based chatbots cannot simulate the pressure of a real job interview.

Our project is a real-time, voice-first training environment where the AI agent acts as a hiring manager. It doesn’t just talk; it actively listens, judges your answers, and controls the UI based on your performance.

The Core Flow:

Setup: The user authenticates via Clerk and pastes their resume and a job description.

Configuration: The user selects a specific Persona (e.g., “Strict Tech Lead” or “Visionary Founder”) and sets a Harshness Slider (0–100). This injects specific behavioral instructions into the agent’s system prompt.

The Interview: As the user answers questions via voice, the agent uses Client-Side Function Calling to evaluate the response in real-time.

Feedback Loop: If the user gives a vague answer, the agent triggers a rateAnswer tool that immediately lowers the visual “Confidence Score” bar on the screen and changes the agent’s verbal tone.

Meeting the Criteria:

Working Prototype: The application runs end-to-end. Users can log in, configure the session, and complete a full voice interview with live state updates.

Technical Complexity: We went beyond simple voice-to-text. We implemented bidirectional control: the user speaks to the AI, and the AI controls the frontend (the score bar and toast notifications) using tool execution.

Innovation: We added gamification elements like the “Harshness Slider” and distinct “Personas,” turning a standard QA session into a dynamic roleplay experience.

Theme Alignment: The entire user experience relies on the Conversational Agent. The agent is not just an interface; it is the core logic driver of the application.

Tech Stack:

Framework: Next.js 14, TypeScript, Tailwind CSS

AI & Voice: ElevenLabs React SDK (Conversational AI), Custom Client Tools

Auth: Clerk

UI: Shadcn/ui (Progress bars, Sliders)

Nothing related to this project, this is brand new just for the hackathon