MockMate AI

Inspiration

The inspiration came from seeing talented peers struggle during placements, not because they lacked technical knowledge, but because of interview anxiety and a lack of real-time practice. Traditional preparation (mock interviews with friends) is often inconsistent. We wanted to build a "Safe-to-Fail" environment that is available 24/7, helping candidates build confidence through high-quality AI interaction.

What it does

MockMate AI is a full-stack interview simulator. It’s not just a chatbot; it’s a coach.

It conducts role-specific interviews (Technical, HR, Behavioral).

It listens to the user (Voice-to-Text) and responds with ultra-low latency.

At the end of the session, it provides a Performance Report, analyzing the user's answers, giving a score, and suggesting specific areas for improvement.

How we built it

We focused on speed and security:

Frontend: Built with HTML5, CSS3, and Vanilla JavaScript for a lightweight and responsive UI.

Backend: A Node.js and Express.js server acts as a secure bridge between the user and the AI.

AI Engine: Integrated Groq Cloud (Llama 3.1) to achieve near-instant response times, which is crucial for a natural conversation flow.

Voice Integration: Used the Web Speech API to allow users to speak their answers, making the experience more realistic.

Challenges we ran into

The biggest hurdle was Security and Latency.

Initially, calling the API from the frontend exposed our API keys. We solved this by building a Backend Proxy in Node.js.

We also faced challenges with Deployment on Render, specifically managing environment variables and port configurations, which we resolved through rigorous debugging of the server logs and refining our package.json scripts.

Accomplishments that we're proud of

Zero Latency: Achieving a response time that feels like talking to a human.

Security: Building a production-ready setup where sensitive keys are 100% protected.

Working Prototype: Moving from an idea to a fully functional, deployed web application on Render within the hackathon timeline.

What we learned

We deep-dived into Prompt Engineering—learning how to make an LLM behave like a strict interviewer versus a helpful coach. We also gained hands-on experience in Full-Stack Deployment, understanding how to bridge the gap between a local development environment and a live cloud server.

What's next for MockMate AI

Video Analysis: Integrating computer vision to analyze facial expressions and eye contact.

Resume Parsing: Allowing users to upload their resumes so the AI can ask personalized questions based on their actual experience.

Company-Specific Modes: Training the model to simulate the interview styles of top-tier companies like Google, Amazon, or TCS.

Built With

• backend-proxy
• css
• css3
• dotenv
• environment-variables
• express.js
• fetch-api
• github
• groq-cloud-api
• html5
• javascript
• llama-3.1
• node.js
• npm
• real-time
• render
• rest-api
• web-speech-api