PrepAI is a comprehensive, production-grade AI career mentorship and technical interview preparation platform. Powered by Google's Gemini 2.5 Flash AI and Cloud Firestore, PrepAI simulates enterprise technical screens, audits resumes against Applicant Tracking Systems (ATS), reviews Big-O algorithmic complexity, and delivers 24/7 executive career coaching.
- Principal Engineer Persona: Simulates behavioral (STAR) and system design screens tailored to specific seniority levels (Intern to Staff/Principal).
- Dynamic Follow-ups: Generates context-aware follow-up prompts challenging edge cases, concurrency models, and scalability trade-offs.
- Instant Verdict Scorecard: Evaluates responses with a 10-point rubric, actionable strengths/improvements breakdown, and exemplary model answers.
- Keyword Gap Matrix: Benchmarks plain-text or uploaded resumes against target job specs (Workday, Greenhouse, Lever heuristics).
- Callback Probability Score: Predicts interview selection rates and identifies high-impact missing keywords.
- Tailored Question Prediction: Automatically forecasts custom interview questions based on your resume's past projects.
- Multi-Language Support: Interactive syntax-highlighted coding environment supporting JavaScript, TypeScript, and Python.
- Real-Time Big-O Analysis: Evaluates time and space complexity instantly using Gemini code analysis engines.
- Principal Code Refactoring: Suggests boundary guard improvements, clean code patterns, and O(1) space optimizations.
- STAR Story Structuring: Helps craft compelling behavioral leadership examples.
- Offer Negotiation Strategy: Provides executive scripts and counter-offer frameworks.
- Firebase Authentication: Seamless candidate account synchronization across devices.
- Cloud Firestore: Encrypted, isolated persistence of interview transcripts, ATS records, and practice sessions.
- Demo Candidate Workspace: Includes sample chart telemetry and pre-injected sessions for instant exploration.
- Frontend: React 19, TypeScript, Vite, Tailwind CSS v4, Lucide Icons, Recharts
- Backend Service: Express.js (Node.js/ESM compiled to standalone CJS bundle)
- AI Engine: Google GenAI SDK (
@google/genaiv2.4) utilizinggemini-2.5-flashwith JSON structured generation schemas - Database: Google Firebase Cloud Firestore & Authentication
- Infrastructure: Google Cloud Run (Containerized ingress routing on Port 3000)
Create a .env file in the project root based on .env.example:
# Required for Gemini AI API calls (Server-Side Only)
GEMINI_API_KEY="AIzaSy..."
# Host service URL injected automatically by Cloud Run / AI Studio
APP_URL="http://localhost:3000"Important
API Key Security: All Gemini API calls are proxied exclusively through backend Express routes (/api/ai/*). Sensitive API keys are never exposed to client browser bundles.
-
Install Dependencies:
npm install
-
Start Full-Stack Dev Server:
npm run dev
Access the live application at
http://localhost:3000. Backend API routes and Vite HMR middleware run concurrently on the single container port.
PrepAI is pre-configured for automated bundling and deployment to Google Cloud Run:
-
Production Build:
npm run build
- Compiles the frontend React SPA into optimized static files inside
dist/. - Bundles the backend Express server into a standalone
dist/server.cjsfile viaesbuild.
- Compiles the frontend React SPA into optimized static files inside
-
Start Production Container:
npm start
Launches the standalone Express production server serving both
/api/*endpoints and SPA fallback assets on0.0.0.0:3000.
- Firebase Security Rules: Enforce strict authenticated user isolation (
resource.data.userId == request.auth.uid). - Sanitization: All user submissions are validated server-side prior to AI prompt injection.