CodePilot AI — AI Coding Interview Preparation Platform

A full-stack platform for coding interview prep: Spring Boot + PostgreSQL backend, HTML/CSS/JS frontend, JWT auth, and Gemini-powered AI features throughout.

Note on database: This project ships configured for PostgreSQL (see application.properties). The pom.xml also includes the MySQL driver, so a MySQL setup is technically possible, but the instructions below focus on Postgres since that's what most free hosting platforms (Render, Railway, Supabase, Neon) provide out of the box — this is the easiest path to a live deployment even if you've only used MySQL before. A quick MySQL↔Postgres cheat-sheet is included at the bottom for anyone coming from MySQL.

Features

• 🔐 Auth: Register, login (JWT + BCrypt), forgot/reset password (token-based)
• 📊 Dashboard: solved/attempted counts, difficulty breakdown (donut chart), daily streak, gamified contest rating, GitHub-style activity heatmap
• 🤖 AI Interviewer: ChatGPT-style chat for concept explanations and Q&A
• 💡 AI Hints: 3 progressive hint levels per question (nudge → approach → pseudocode) — never gives the full solution
• 📚 44 practice questions across Arrays, Strings, Linked List, Trees, Graphs, DP, Stack, Heap, Searching, Bit Manipulation
• 🏢 Company-wise questions: browse by Amazon, Google, Meta, Netflix, Apple, Microsoft, Uber, Adobe, Oracle, Bloomberg
• 📝 Notes: write and auto-save personal notes per question
• ⭐ Bookmarks: save favorite questions
• 📄 Resume Analyzer: upload a PDF, get an AI-generated ATS score, weak points, missing keywords, and detected skills
• ⏱ Mock Interview: 5 random questions, countdown timer per question, end-of-round score
• ⚙️ Admin Panel: add/edit/delete questions, view all registered users (protected by ROLE_ADMIN)
• Dark, glassmorphic UI with subtle animations
• Swagger/OpenAPI docs at /swagger-ui.html

Tech Stack

Backend: Java 17, Spring Boot 3.5, Spring Data JPA, Spring Security, JWT (jjwt), PostgreSQL, Apache PDFBox, springdoc-openapi Frontend: HTML, CSS, vanilla JavaScript (served as static files by Spring Boot) AI: Gemini API (default) or OpenAI API

---

1. Set up PostgreSQL

You have two easy options — pick whichever is less friction for you.

Option A: Install PostgreSQL locally (for local development)

Windows/Mac: download the installer from postgresql.org/download and run it (it installs a GUI tool called pgAdmin too, which works a lot like MySQL Workbench).

Mac (Homebrew):

brew install postgresql@16
brew services start postgresql@16

Linux:

sudo apt install postgresql postgresql-contrib
sudo service postgresql start

Once installed, create the database:

# Open the Postgres shell (equivalent of "mysql -u root -p")
psql -U postgres

CREATE DATABASE codepilotdb;

Type \q to exit the shell.

Option B: Use a free hosted PostgreSQL (recommended for deployment)

Since you're deploying (not just running locally), a managed Postgres instance saves you from installing anything:

• Render – free Postgres instance, pairs naturally with a Render web service for the Spring Boot app
• Railway – one-click Postgres, gives you a connection URL instantly
• Supabase – free Postgres with a nice dashboard
• Neon – free serverless Postgres

Any of these will hand you four values you need in step 2: host, port, database name, username, password (or one combined connection URL).

---

2. Configure the database connection

The app already reads these as environment variables (with local defaults baked in), so you don't need to edit application.properties at all — just set env vars:

Variable	Example (local)	Example (hosted)
SPRING_DATASOURCE_URL	jdbc:postgresql://localhost:5432/codepilotdb	jdbc:postgresql://<host>:5432/<dbname>
SPRING_DATASOURCE_USERNAME	postgres	(given by your host)
SPRING_DATASOURCE_PASSWORD	(your local password)	(given by your host)
JWT_SECRET	any long random string	same
GEMINI_API_KEY	your key from aistudio.google.com/app/apikey	same

Locally, you can set these in your terminal before running, e.g.:

export SPRING_DATASOURCE_URL=jdbc:postgresql://localhost:5432/codepilotdb
export SPRING_DATASOURCE_USERNAME=postgres
export SPRING_DATASOURCE_PASSWORD=your_local_password
export JWT_SECRET=change_this_to_a_long_random_string
export GEMINI_API_KEY=your_gemini_key

On your host (Render/Railway/etc.), set the same names in the service's "Environment Variables" panel — most hosted Postgres providers give you a ready-made SPRING_DATASOURCE_URL-style connection string you can paste directly.

If you'd rather hardcode values for a quick local test, you can still edit them directly in src/main/resources/application.properties:

spring.datasource.url=jdbc:postgresql://localhost:5432/codepilotdb
spring.datasource.username=postgres
spring.datasource.password=YOUR_POSTGRES_PASSWORD

No manual table creation needed — spring.jpa.hibernate.ddl-auto=update means Hibernate creates/updates all tables automatically on startup.

---

3. Run the backend

./mvnw spring-boot:run

Runs on http://localhost:8080. On first startup:

• 44 practice questions are seeded automatically (with company tags)
• A default admin account is created: admin@codepilot.ai / Admin@123 — change this password in any real deployment, and use it to access /admin.html

4. Use the app

Open http://localhost:8080 — you'll be routed to login/register automatically.

Page	Path
Register	/register.html
Login	/login.html
Forgot password	/forgot-password.html
Dashboard	/dashboard.html
Practice Questions	/questions.html → click into /question.html?id=N
Companies	/companies.html
AI Interviewer	/mentor.html
Resume Analyzer	/resume.html
Mock Interview	/mock-interview.html
Admin Panel	/admin.html (ADMIN role only)

---

Deploying (Docker)

A Dockerfile is included and builds/runs the app out of the box:

docker build -t codepilot-ai .
docker run -p 8080:8080 \
  -e SPRING_DATASOURCE_URL=jdbc:postgresql://<host>:5432/<dbname> \
  -e SPRING_DATASOURCE_USERNAME=<user> \
  -e SPRING_DATASOURCE_PASSWORD=<password> \
  -e JWT_SECRET=<your_secret> \
  -e GEMINI_API_KEY=<your_key> \
  codepilot-ai

Point SPRING_DATASOURCE_URL at your hosted Postgres instance from step 1B and it just works — no code changes needed.

---

Coming from MySQL? Quick cheat-sheet

What you know (MySQL)	Postgres equivalent
mysql -u root -p	psql -U postgres
SHOW DATABASES;	\l
USE mydb;	\c mydb
SHOW TABLES;	\dt
DESCRIBE table_name;	\d table_name
MySQL Workbench (GUI)	pgAdmin (GUI, installed alongside Postgres) or TablePlus
Default port 3306	Default port 5432
Default admin user root	Default admin user postgres
AUTO_INCREMENT	SERIAL / GENERATED ALWAYS AS IDENTITY (Hibernate handles this for you automatically — no action needed)

You don't need to write any raw SQL yourself for this project — Hibernate/JPA generates and manages all the tables. You only need psql/pgAdmin to create the initial empty database and, optionally, to peek at the data.

---

API Overview

Method	Endpoint	Auth	Description
POST	/api/auth/register	No	Create account
POST	/api/auth/login	No	Log in
POST	/api/auth/forgot-password	No	Get a reset token
POST	/api/auth/reset-password	No	Reset password with token
GET	/api/questions/all	No	List all questions
GET	/api/questions/{id}	No	Get one question
GET	/api/questions/topic/{topic}	No	Filter by topic
GET	/api/questions/difficulty/{difficulty}	No	Filter by difficulty
POST/PUT/DELETE	/api/questions/...	ADMIN	Manage questions
POST	/api/mentor/ask	Yes	Chat with AI Interviewer
POST	/api/mentor/hint	Yes	Get a progressive hint (level 1-3)
GET/POST	/api/progress, /api/progress/update	Yes	Dashboard stats
GET	/api/progress/heatmap	Yes	Activity heatmap data
GET/POST/DELETE	/api/bookmarks	Yes	Manage bookmarks
GET/PUT	/api/notes/{questionId}	Yes	Per-question notes
POST	/api/resume/analyze	Yes	Upload + analyze resume PDF
GET	/api/admin/users	ADMIN	List all users

Full interactive docs: http://localhost:8080/swagger-ui.html

Project structure

codepilot-ai/
├── pom.xml
├── Dockerfile
└── src/main/
    ├── java/CodePilotAI/
    │   ├── config/     (Security, CORS, Swagger, WebClient, DataSeeder, AdminSeeder)
    │   ├── controller/ (Auth, Question, Mentor, Bookmark, Progress, Note, Resume, Admin)
    │   ├── dto/
    │   ├── entity/     (User, Question, Bookmark, StudentProgress, Note, ActivityLog, PasswordResetToken)
    │   ├── exception/
    │   ├── repository/
    │   ├── security/   (JWT service, filter, UserDetailsService)
    │   └── service/
    └── resources/
        ├── application.properties
        └── static/     (13 pages — see table above)

Design notes & honest limitations

• Forgot password works end-to-end (token generation, expiry, reset), but since there's no SMTP server configured, the reset token is returned directly in the API response and shown on-screen instead of emailed. Wire up a mail sender in UserService if you need real email delivery.
• Contest rating is an internal gamified score (+5/+12/+25 per easy/medium/hard solve), not a true competitive rating system — there's no real contest infrastructure behind it.
• Company tags are topic-based groupings for practice variety, not a verified "this exact question was asked at this company" dataset.
• No real code execution/judging — there's a code editor to draft your solution and a "Mark Solved/Attempted" flow that updates your dashboard, plus a direct link to the real LeetCode problem to verify. Building an actual sandboxed code judge (running arbitrary user code safely) is a separate, much larger project with real security considerations.
• Resume ATS scoring is AI-generated (Gemini) based on resume text extracted via PDFBox — treat scores as directional feedback, not a certified ATS simulation.