Smart Interview Coach

Stack: MERN (MongoDB, Express, React, Node) • Cloud: AWS Lambda + API Gateway + S3 • AI: OpenAI API

An AI‑driven full‑stack platform that analyzes resumes and generates personalized interview questions & feedback. Designed to be simple to run locally and easily deployable with Serverless to AWS Lambda.

---

🔭 Goals (Scope)

• Upload → Parse → Analyze → Feedback pipeline with minimal moving parts
• Serverless backend (Express on Lambda) to keep ops simple and cheap (free‑tier friendly)
• AI integration via OpenAI (abstracted behind a service; swap with local LLM if desired)
• S3 storage for original uploads and parsed text (optional; local disk when developing)
• Clean API and clear code flow

---

🧱 Architecture (High Level)

React (Vite) ──────> API Gateway ──────> Lambda (Express via serverless-http)
     │                         │
     │                         └──────> S3 (uploads)
     │
     └────────────────────────────────> MongoDB Atlas (metadata, sessions)
                              
Lambda ──> OpenAI (analysis) ──> returns fit-scores, questions, feedback

• Frontend: React app for upload, status, results
• Backend: Express APIs (/analyze/upload, /analyze/status/:id, /analyze/result/:id)
• AI service: Small wrapper calling OpenAI with guardrails & cost controls
• Storage: MongoDB Atlas (free tier) for metadata; S3 (optional) for files

See docs/ARCHITECTURE.md for sequence diagrams & code flow.

---

🚀 Quickstart (Local)

1) Prereqs

• Node 20+ and npm (or pnpm)
• MongoDB Atlas or local MongoDB (optional for first run)
• OpenAI API Key (or set USE_MOCK_AI=true)

2) Backend

cd backend
cp .env.example .env
# edit .env to add OPENAI_API_KEY (or USE_MOCK_AI=true), and AWS creds if using S3 locally
npm install
npm run dev

3) Frontend

cd frontend
npm install
npm run dev
# open http://localhost:5173

---

☁️ Deploy to AWS (Serverless Framework)

Setup

npm install -g serverless
cd backend
npm install
serverless deploy

This provisions:

• Lambda function (Node 20)
• API Gateway endpoint
• IAM role with minimal S3 access (if enabled)

Get the endpoint from the deploy output and set VITE_API_BASE in frontend/.env.

---

🔐 Environment Variables

Backend (backend/.env):

# OpenAI
OPENAI_API_KEY=
USE_MOCK_AI=false

# Mongo
MONGO_URI=

# AWS (optional for S3 usage locally)
AWS_REGION=us-east-1
S3_BUCKET=smart-interview-coach-dev

# App
PORT=8080
ALLOW_ORIGIN=http://localhost:5173

Frontend (frontend/.env):

VITE_API_BASE=http://localhost:8080

---

🧪 API (Essential)

• POST /analyze/upload — multipart/form-data (file), returns { jobId }
• GET /analyze/status/:jobId — returns { status }
• GET /analyze/result/:jobId — returns { resumeText, questions[], feedback, fitScores{} }

Detailed docs: docs/API.md

---

🧭 Code Flow

1. Upload (React → /analyze/upload) → backend stores file (local or S3), extracts text, creates jobId
2. Analyze → AI service (OpenAI or mock) generates questions & feedback
3. Poll status → retrieve results when ready
4. Display dashboard (summary + download JSON)

Sequence diagrams & notes: docs/ARCHITECTURE.md

---

🧰 Scripts

• npm run dev (backend) — nodemon-based local server
• npm run deploy (backend) — serverless deploy shortcut
• npm run dev (frontend) — Vite dev server

---

📝 License

MIT — see LICENSE