Aiverse

ML Engineering Platform for Serious Developers

Train. Compete. Ship ML Systems.

Live → aiverse-webai.vercel.app

---

What is Aiverse?

Most ML platforms are either theory-heavy courses or competition-focused arenas like Kaggle. Neither teaches you to actually build and ship ML systems.

Aiverse sits in between:

Learn → Experiment → Solve → Improve → Repeat

One platform. Real execution. No fake outputs.

---

Features

ML Problems

Solve real ML challenges with actual datasets. Write code in the browser, submit, and get evaluated by a sandboxed backend executor — similar to LeetCode but for machine learning.

• Classification, regression, clustering problems
• In-browser code editor
• Docker-sandboxed execution
• Instant feedback and scoring
• ELO-based leaderboard ranking

ML Playground

An interactive experimentation lab. No setup required.

1. Pick a dataset
2. Choose a model (Logistic Regression, Random Forest, SVM, KNN)
3. Tune hyperparameters
4. Train
5. Visualize accuracy, loss, confusion matrix

Designed for exploration, not just submission.

AI Mentor

Ask any ML-related question and get a contextual explanation — not a generic answer. Currently improving personalization and user-awareness.

Learn

Structured ML curriculum covering:

• Classification and regression
• Model evaluation metrics
• Overfitting and regularization
• Core algorithms with examples

Discussions

Community Q&A. Share approaches, ask doubts, learn from others.

---

Tech Stack

Layer	Technology
Frontend	React 18, Tailwind CSS, Framer Motion
Backend	Django 4.2, Django REST Framework
Database	PostgreSQL
Async	Celery + Redis
Sandbox	Docker (isolated code execution)
ML	scikit-learn, pandas, NumPy
Deploy	Vercel (frontend), Render (backend)

---

Getting Started

Prerequisites

• Python 3.10+
• Node.js 18+
• Docker (for sandbox + Redis)
• PostgreSQL

1. Clone the repo

git clone https://github.com/deepmhatre13/Aiverse.git
cd Aiverse

2. Backend setup

cd aiverse_backend/backend
python -m venv venv

# Windows
venv\Scripts\activate

# Mac/Linux
source venv/bin/activate

pip install -r requirements.txt
cp .env.example .env   # fill in your values
python manage.py migrate

3. Start services

# Redis (via Docker)
docker run -p 6379:6379 redis

# Celery worker
celery -A backend worker --loglevel=info --pool=solo

# Django
python manage.py runserver

4. Frontend setup

cd aiverse_frontend
npm install
npm run dev

Open http://localhost:5173

---

Environment Variables

Create a .env file in aiverse_backend/backend/:

SECRET_KEY=your-django-secret-key
DEBUG=True
DATABASE_URL=postgresql://postgres:password@localhost:5432/aiverse_db
REDIS_URL=redis://localhost:6379/0

---

---

What Makes It Different

Typical Student Projects	Aiverse
Static UI demos	Real backend execution
Hardcoded ML outputs	Live pipeline runs in Docker
Disconnected modules	Integrated learn → solve → improve loop
No evaluation logic	Actual scoring against ground truth

---

Current Limitations

Being honest about where it stands:

• Mentor responses are not deeply personalized yet
• Playground uses static datasets
• No experiment history or comparison across runs
• Modules are loosely connected (improving)

---

Roadmap

• User intelligence layer — track patterns, surface insights
• Experiment history and side-by-side comparison
• Personalized problem recommendations
• Smarter mentor with user context awareness
• More problem types (NLP, time series, CV)

---

Deployment

Service	Platform
Frontend	Vercel
Backend	Render
Database	Render PostgreSQL
Redis	Render Redis

---