Evalvate

Before interviews, candidates wonder am I speaking too fast, do I sound confident? The problem is, they never know. Evalvate solves that gap

Project Name

Evalvate - AI Interview Performance Coach

Demo

Live Demo: Check out the frontend of Evalvate

GitHub Repository

Source Code: The backend and frontend source Code

Project Description

Inspiration

Candidates constantly wonder: “Am I speaking too fast? Am I confident enough? Should I smile more?”

Yet they rarely get real answers. 48% of students fail interviews not due to lack of skills, but lack of confidence, clarity, and feedback. Universities struggle with placement outcomes; companies face costly mis-hires.

Evalvate was built to fix this feedback gap.

What It Does

Evalvate is an AI-powered interview coach that delivers structured and personalized feedback.

Key Capabilities:

AI-simulated interviews tailored to roles & industries
Voice, text & face-analysis for interview behavior
Metrics on clarity, tone, confidence, pacing, filler words
Real-time feedback + improvement suggestions
Progress dashboards to track growth

Users:

Students
Universities
Corporates

How We Built It

Frontend: HTML, CSS, JS, WebRTC for video capture
Backend: Node.js, Prisma, PostgreSQL
AI/ML : Gemini + huggingFace Models (audio tone, facial expression, text scoring)
- Cloud: Vercel deployment

System Flow: Video/Audio/Text -> ML Analysis -> Gemini Scoring -> Feedback Report

How We Used the Gemini API

Evalvate integrates Gemini 1.5 Flash & Gemini Pro for:

Use Case	Gemini Role
Answer evaluation	Scoring clarity, structure, completeness
Behavioral feedback	Tone + confidence interpretation
Interview generation	Dynamic interview questions per role/domain
Improvement suggestions	Personalized coaching recommendations
Transcript refinement	Audio transcription -> structured text for analysis

Example: We send the transcript + metadata (tone, pace, pauses) to Gemini to generate:

Score breakdown
Strengths & weaknesses
Improvement tips
Example ideal answers

This turns raw interview data into professional-quality coaching feedback.

Challenges

Making feedback objective, not generic
Handling accents & varied speaking styles
Efficient ML inference (video/audio processing)
UX that reduces anxiety, not increases it
Pricing fairly for students while sustainable for scaling

Accomplishments

AI prototype analyzing real interview responses
Early pilot interest from universities
We use Evalvate ourselves improving our own interview confidence

What We Learned

Interview prep isn’t just about “what” you say, but how you say it. Students crave specific, actionable feedback, not generic “good/bad.” Universities want measurable insights, not vague placement reports. Building trust with users requires empathy, we tested Evalvate on ourselves first.

What's Next

Scale pilots across universities in the next 6 months. Expand to corporates to cut down on costly mis-hires. Build multilingual support for diverse student populations. Extend beyond interviews into admissions, leadership training, and public speaking. Ultimately, make Evalvate the go-to platform for feedback anytime performance matters.

Vision: Become the global platform for AI-powered performance feedback.