What I Built
I built Interview Buddy AI, a local AI-powered mock interview coach for a friend who is preparing for job interviews.
The problem was simple: practicing interviews alone can be difficult. My friend needed a way to practice answering interview questions, receive feedback, and improve without needing another person to conduct every practice interview.
Interview Buddy AI provides a mock interview experience where the AI asks interview questions, evaluates answers, gives a score and feedback, and can ask follow-up questions.
It supports both text and voice-based interview practice.
Demo
GitHub:
https://github.com/ashandeepkaur27/Interview_Buddy-AI
Video Demo:
The demo shows:
Starting an interview
AI-generated interview questions
Answering questions
AI feedback and scoring
Follow-up questions
Voice input and transcription
Code
The complete source code is available here:
https://github.com/ashandeepkaur27/Interview_Buddy-AI
The repository includes the application code, README, requirements, setup instructions, .gitignore, and MIT License.
How I Built It
Interview Buddy AI is built around Gemma 3, an open-weight AI model running locally through Ollama.
Gemma 3
I use Gemma 3 to generate realistic interview questions, evaluate answers, provide feedback, and generate follow-up questions.
Ollama
I use Ollama to run Gemma 3 locally instead of depending on a cloud AI API for the core interview functionality.
Faster-Whisper
For voice-based interviews, Faster-Whisper converts spoken answers into text locally so the AI can evaluate them.
Streamlit
The application interface is built with Streamlit.
The overall flow is:
User → Interview Buddy AI → Gemma 3 through Ollama → Question / Evaluation / Feedback
For voice:
User's Voice → Faster-Whisper → Text → Gemma 3 → Feedback
Why Does Open Innovation Matter?
Open innovation made this project possible without depending completely on a closed AI API.
Running Gemma 3 through Ollama allows the core AI experience to run locally on the user's own computer. This gives users more control over their data and setup.
It also makes experimentation easier. Developers can change models, experiment with prompts, and modify the application's behaviour without being locked into a single closed AI provider.
For an interview practice tool, this is useful because users may share personal information about their education, projects, career plans, and previous interview experiences.
The open approach allowed me to build a more customizable and locally focused interview practice experience.
What I Learned
Building this project helped me understand how local AI models can be integrated into a real application.
I learned how to connect an application with Ollama, work with an open-weight model such as Gemma 3, process voice input with Faster-Whisper, and combine these components into a practical AI application.
My biggest takeaway was that AI applications do not always need to depend on a closed cloud API. Local models can be useful when privacy, customization, and control are important.
Prize Categories
Best Use of Gemma
Interview Buddy AI uses Gemma 3 as the core AI model for generating interview questions, evaluating answers, providing feedback, scoring responses, and generating follow-up questions.
Top comments (0)