Scribble Pad at IIT-B - AI Edition 💡

Participate in Intense Hackathon in JavaScript for AI from 5th February to 16th February, 2025

Table of Contents

Are you ready to push the boundaries of AI, WebAssembly, and Scientific Computing? The Scribble Pad Hackathon challenges you to build groundbreaking deep-tech solutions entirely in JavaScript and WebAssembly. From tree ensemble learning models to LLMs in the browser, from Stable Diffusion frameworks to high-performance scientific computing, this is your chance to work on cutting-edge technologies. Collaborate with peers, build open-source tools, and showcase your innovations in a Scribbler Notebook. With prize money, certificates, and networking opportunities, this is the perfect stage to learn, create, and compete!

Form for Registration #

Problem Statements #

Problem 1: Framework for AI Agents in JavaScript: #

Goal: Create an easy-to-use framework for building AI agents in JavaScript, focusing on decision-making, learning, and interaction. Techniques:

  • Modular Components: For decision-making, learning, and interaction.
  • Reinforcement Learning: Methods like Q-Learning for autonomy.
  • LLM Integration: For chatbots and voice assistants.
  • Multi-agent Systems: Collaboration/competition among agents.
  • Simulation Tools: For environment interaction and testing.

Problem 2: Tree Ensemble Learning Model (like LightGBM or Catboost) in JavaScript: #

Goal: Create a tree-based machine learning model that can be used both in Node.js and the browser. It should work efficiently and can be written directly in JavaScript or via a C implementation ported through WebAssembly (WASM). Techniques: Participants could explore:

  • Decision Trees and Gradient Boosting for implementing ensemble models.
  • WASM for optimizing performance (potentially using libraries like emscripten to port C++ code).
  • WebGPU for leveraging hardware acceleration.
  • Libraries/Tools: TensorFlow.js, Brain.js, or directly manipulating decision trees with JavaScript.

Problem 3: Voice Gen in the Browser: #

Goal: Build a library to load and run multi-model voice-gen in the browser using WASM and WebGPU for performance optimization. Techniques: Focus on:

  • Model Quantization to reduce model size for the browser.
  • WebAssembly (WASM) for the core computation.
  • WebGPU for accelerating model inference on GPUs.
  • Participants could explore existing libraries like webgpu and onnxruntime-web for loading models in the browser.

Problem 4: Library for Doing Scientific Computing (Bioinformatics/CFD/Electromagnetic Simulations/Mechanical Simulations): #

Goal: Build a library in JavaScript that handles advanced scientific computations, particularly in fields like bioinformatics, CFD (Computational Fluid Dynamics), electromagnetics, and mechanical simulations. Techniques:

  • Implement mathematical solvers (Finite Element Method, Fast Fourier Transform, etc.).
  • Use WebAssembly for intensive computations, especially for fluid dynamics, electromagnetics, and simulations.
  • Explore JavaScript libraries like math.js or custom solutions for matrix operations, differential equations, and data manipulation.

📅 Timeline #

February 5th – Kickoff Workshop #

  • Introduction to problem statements, technologies, and expectations.
  • Hands-on session to help participants get started.

    February 5th - 8th – Registration Period #

  • Last date for students to register: February 8th
  • February 8th - 15th – Offline Hackathon
  • Teams work on their projects, build prototypes, and refine their solutions.

    February 16th Morning – Finalists Announcement 🎉 #

  • Top teams shortlisted for the final demos.

    February 16th Afternoon – Demos, Awards & Networking #

  • Finalists present their projects.
  • Winners announced, awards distributed.
  • Networking session with industry experts, mentors, and fellow participants.

Benefits for Participants 🚀🎯 #

1️⃣ Prize Money 🏆💰 #

Compete for exciting cash rewards for the best projects!

2️⃣ Demoable Open-Source Output #

Build cutting-edge projects that you can showcase in your portfolio, GitHub, or future applications.

3️⃣ Certificates of Participation & Achievement #

Receive official certificates to add to your resume and LinkedIn profile.

4️⃣ Joy of Learning & Teamwork #

Collaborate with peers, explore deep tech, and gain hands-on experience in cutting-edge fields like AI, WASM, and WebGPU.

Hackathon Rules & Guidelines #

1️⃣ Code must be in JavaScript #

Core implementation should be in JavaScript. C/C++ are allowed only if compiled to WebAssembly (WASM).

2️⃣ Deep Tech Focus #

Projects must align with one of the provided problem statements (tree ensemble models, LLMs in browser, Stable Diffusion, or scientific computing).

3️⃣ ES6 Compatibility #

The final module must be an ES6 module that can be loaded directly in the browser.

4️⃣ Scribbler Notebook for UI #

The final UI should be presented in a Scribbler notebook. Teams can host scripts on GitHub and import them into Scribbler.

5️⃣ Judges’ Decision is Final ⚖ #

Submissions will be evaluated based on technical depth, innovation, and execution. The judges’ decision is final and cannot be contested.

Use Scribbler - Join Discord - Checkout Hub - Star on Githib