lishuangqiang/AI-Meeting

What it solves

This project provides an AI-powered mock interview and resume analysis platform. It solves the problem of preparing for technical interviews by automating the process of resume parsing, generating personalized interview questions based on a candidate's experience, and simulating a multi-stage interview process with real-time feedback.

How it works

The system uses a modular monolithic architecture built with Spring Boot and Spring AI. It integrates several key components:

• Resume Analysis: An AI agent parses uploaded PDF resumes to generate structured interview questions and initial scores.
• Multi-Agent Orchestration: Five specialized agents (Questioner, Interviewer, Scorer, Follow-up, and Expression Analyst) collaborate to conduct the interview. A rule engine (LiteFlow) manages the logic for follow-up questions.
• Real-time Interaction: It uses WebSockets for real-time speech-to-text (ASR) and Server-Sent Events (SSE) for streaming AI text responses.
• State Management: To prevent data loss during long sessions, it employs a tiered snapshot system using MongoDB and Redis.
• Optimization: A distributed "Single-flight" framework ensures that identical AI requests across multiple server instances are only executed once to reduce costs and latency.

Who it’s for

• Job seekers looking to practice technical interviews with AI-driven, personalized questioning.
• Developers wanting to study a production-ready implementation of AI agents, real-time voice processing, and distributed system patterns in Java.
• Students seeking a comprehensive project for a graduation thesis or portfolio.

Highlights

• Resume-Driven Questioning: Generates questions deeply linked to the candidate's specific professional experience.
• Multi-Agent Collaboration: Simulates a realistic interview with specialized roles for scoring, questioning, and behavioral analysis.
• Real-time Voice Integration: Features end-to-end real-time speech transcription and long-text speech synthesis.
• Expression Analysis: Evaluates the candidate's demeanor via camera screenshots using an AI expression recognition agent.
• Comprehensive Feedback: Provides a post-interview report featuring a multi-dimensional radar chart and detailed question-by-question analysis.