π‘ Inspiration
Interview prep is stressful β juggling multiple rounds, guessing what questions might come up, and trying to assess your own performance without any real feedback. Traditional mock interviews are expensive, time-consuming, and rarely personalized.
Our goal was to create something that feels like having a personal mentor available anytime β one that knows your resume, understands the kind of role youβre targeting, and can coach you with empathy and intelligence.
Thatβs how Interview GenAIe β your AI-powered interview coach β was born.
π What it does
Interview GenAIe is an intelligent, voice-first interview simulator that tailors every question to your resume, chosen interview round, and career goals.
It allows users to:
- π― Choose an interview type: HR, Technical, Managerial, or General.
- π§Ύ Upload a resume: the system analyzes it and generates context-aware, role-specific questions.
- π§ Experience lifelike conversations: powered by ElevenLabsβ ultra-realistic voice synthesis, Interview GenAIe speaks like a real interviewer, bringing empathy and presence into every mock interview.
- π¬ Receive instant AI feedback: detailed analysis on clarity, confidence, tone, and content quality.
- π§ Get suggested model answers: see how to improve and iterate smarter.
- π Use non-voice mode: if users prefer, they can simply generate questions and receive AI-generated sample answers directly β without recording responses.
- β Answer rating system: every answer is scored and explained, showing users which specific areas need improvement (e.g., structure, depth, tone).
In short, Interview GenAIe helps users practice, improve, and gain confidence β anytime, anywhere β making interview prep interactive, data-driven, and deeply personal.
π§© How we built it
Interview GenAIe combines modular AI services with a user-friendly Streamlit front-end for a seamless, voice-based experience.
π» Frontend
- Built entirely in Streamlit, offering a responsive, voice-first interface.
- Manages user sessions, resume uploads, audio recording, and round selection through
app.py.
βοΈ Core Logic & AI Services
- LangChain + OpenAI APIs: dynamic question generation, resume-based context retrieval, and AI feedback.
- Custom Prompt Modules:
question_prompts.pyandfeedback_prompts.pygenerate round-specific, structured responses. - Feedback Engine:
feedback_generator.pyevaluates user performance and produces detailed improvement suggestions.
π§ Audio Pipeline
- π£οΈ ElevenLabs powers the voice of Interview GenAIe β transforming text-based questions into expressive, human-like dialogue. It brings warmth, tone, and realism that traditional TTS systems lack, making the mock interview feel natural and immersive.
- π Speech-to-Text: Custom pipeline using Streamlit audio recorder and OpenAI transcription API, allowing users to answer questions verbally and view transcripts instantly.
- Audio Management: Handled by
audio_io.py, ensuring smooth playback and conversion across rounds.
π Resume Parsing
- Implemented in
resume_parser.pyusing PyMuPDF, pypdf, and docx2txt to extract relevant skills and experience for contextualized question generation.
π Session Management
round_manager.pymanages multi-round interview flow and stateful interactions.interview_agent.pyorchestrates the overall logic across modules.
βοΈ Deployment
- Deployed on Streamlit Community Cloud, ensuring zero-setup accessibility and fast iteration.
π§ Model Stack
- π§© LLM: GPT-4o mini (via LangChain)
- π§ Text-to-Speech: ElevenLabs (ultra-realistic, human-sounding interviewer voice)
- π£οΈ Speech-to-Text: Custom transcription pipeline (OpenAI + Streamlit Audio Recorder)
- π₯οΈ Framework: Streamlit
- π Resume Parsing: PyMuPDF, docx2txt, pypdf
- π¬ Feedback Engine: Custom prompt templates + scikit-learn utilities
- βοΈ Deployment: Streamlit Community Cloud
βοΈ Challenges we ran into
- Implementing seamless audio recording within Streamlit required extensive exploration, as the framework doesnβt natively support advanced audio handling. We experimented with multiple community components and custom integrations before achieving a smooth, real-time recording experience.
- Ensuring accurate and fast transcription while keeping latency low during live sessions.
- Making the feedback tone sound human and constructive instead of purely algorithmic.
- Handling resume-parsing edge cases β ensuring the model accurately understands varied resume formats and skills.
π Accomplishments that weβre proud of
- Developed a dual-mode interview system β users can now choose between interactive voice interviews or instant question-answer learning sessions.
- Integrated ElevenLabsβ voice synthesis to deliver one of the most natural, engaging mock interview experiences possible β bridging emotion and intelligence.
- Built a modular architecture with clear separation of logic β parsing, audio, feedback, and LLM orchestration.
- Designed and deployed an end-to-end AI system on Streamlit Cloud that anyone can access instantly.
π What we learned
- Extending Streamlit for real-time voice interaction is possible with creativity β custom components and async workflows are key.
- ElevenLabs voices dramatically enhance user engagement and realism in AI-based interactions.
- LangChain prompt tuning and using GPT-4o mini allows for fast, high-quality responses that maintain contextual accuracy.
- Empathetic AI feedback is critical β human-sounding voice + balanced phrasing encourages user confidence.
- Building modularly (separate parsing, audio, LLM layers) ensures scalability for future enterprise or educational use.
π Whatβs next for Interview GenAIe
- π Progress Dashboard β visualize user improvement across communication, technical accuracy, and confidence.
- π― Adaptive Follow-up Engine β use performance metrics to generate progressive difficulty levels and deeper questions.
- π§ Expanded Feedback Metrics β measure tone, pacing, filler words, and emotional expression for richer analytics.
- π± Mobile App Version β enable quick practice sessions via voice on Android and iOS.
- π University & Career Platform Partnerships β integrate Interview GenAIe into placement and training programs.
Built With
- docx2txt
- elevenlabs
- gpt-4o
- langchain
- openai
- prompt
- pymupdf
- pypdf
- python
- streamlit
- streamlitcommunitycloud
- whisper

Log in or sign up for Devpost to join the conversation.