
Our project reimagines technical hiring by replacing LeetCode-style puzzles with a fully customized, holistic AI evaluation pipeline. A candidate begins by uploading their resume and task solution. Gemini analyzes the submitted code, generating a functionality-based quality score, a natural-language description of what the code achieves, and a set of personalized questions for a dynamic AI-driven video interview. The candidate completes this custom interview while the system records responses, evaluates communication, and administers deterministic MCQ questions with camera-on monitoring. Qdrant stores the task description and code output to compute a code similarity fit, while separately storing the resume to evaluate alignment with an ideal candidate profile. Gemini then synthesizes all signals, including code quality, resume fit, code–task similarity, MCQ performance, and interview transcription, to generate structured feedback and a final weighted score. The result is a fair, inclusive, real-world hiring experience that evaluates the whole candidate rather than their puzzle-solving ability.
19 Nov 2025