Inspiration
We noticed that most students struggle to practice interviews effectively. Current tools are either static or don’t feel like real conversations. We wanted to create something that simulates a real interviewer who actually listens and responds.
What it does
We built an AI-powered mock interviewer that allows students to practice interviews through both text and voice. The AI asks questions, analyzes responses, and generates intelligent follow-up questions in real time, creating a realistic interview experience.
How we built it
We used OpenRouter as the AI layer to process responses and generate follow up questions. On the frontend, we integrated browser-based speech recognition to convert voice to text and speech synthesis to allow the AI to speak back, enabling full voice interaction.
Challenges we ran into
We learned how to design AI systems that feel conversational and natural, not robotic. We also gained experience integrating speech technologies and managing real time interactions between users and AI.
Accomplishments that we're proud of
We built a fully working AI interviewer that supports both text and voice interactions. The system can listen to answers, understand context, and ask relevant follow-up questions in real time, creating a realistic interview experience.
What we learned
We learned how to design AI systems that feel natural and conversational, not robotic. We also gained experience working with speech recognition, handling API limits, and optimizing prompts to get better AI responses.
What's next for MockMate
Next, we plan to improve the voice experience to feel more real-time and seamless, including faster responses and smoother interactions. We also want to personalize interviews based on specific roles or job descriptions, and expand features like saving sessions, tracking progress over time, and scaling to more users.
Built With
- css
- google-oauth
- gpt-4o
- next.js-15
- openrouter
- postgresql
- react-19
- supabase
- tailwind
- typescript
