Welcome to the official GitHub organization of the ACM Student Chapter, IIT Mandi.

We are a community of students interested in computer science, software development, research, and understanding how things work beyond the abstractions we normally use.

ACM Code Labs is a structured initiative by the ACM Student Chapter, IIT Mandi that focuses on learning through implementation.

The idea is straightforward. We start with fundamental concepts, implement them ourselves, understand the details that are usually hidden behind libraries and frameworks, and eventually move towards implementing ideas from research papers.

Fundamentals
     ↓
Implement from Scratch
     ↓
Build Technical Intuition
     ↓
Read Research Papers
     ↓
Implement and Experiment

The goal is not to avoid libraries forever. The goal is to understand what they are doing before using them as a black box.

The initiative is broadly divided into two phases.

We begin with fundamental concepts and implement them without relying heavily on high level abstractions.

Some of the areas covered include:

• Machine Learning Foundations: Linear regression, logistic regression, gradient descent, optimization, and related concepts.
• Modern Architectures: Attention mechanisms, positional encodings, transformers, and the ideas behind modern neural architectures.
• Software Engineering: Modular code, testing, documentation, debugging, and reproducible experiments.

The emphasis is on understanding both the theory and the implementation. If something works, we should ideally know why it works.

After building a strong foundation, members move towards implementing concepts from research papers.

This involves:

• Reading and breaking down research papers.
• Understanding the problem, assumptions, mathematics, and proposed approach.
• Converting equations, architectures, and pseudocode into working implementations.
• Reproducing experiments where practical.
• Documenting implementation choices and results.
• Exploring areas such as deep learning, distributed systems, computer vision, and other fields of computer science.

Reading a paper is one thing. Getting its implementation to work is usually where the interesting part begins.

We encourage members to learn by building, reviewing each other's work, and contributing to ongoing projects.

A typical workflow is:

1. Choose a project or topic from the learning roadmap.
2. Read the relevant material and understand what you are trying to implement.
3. Implement and experiment with the concept.
4. Test and document your work properly.
5. Open a Pull Request and get your implementation reviewed.

You do not need to know everything before starting. That would make this initiative considerably less useful.

ACM Code Labs is ultimately about developing the ability to:

• Understand concepts beyond their high level APIs.
• Translate theory and mathematics into working code.
• Read and interpret technical research.
• Design and implement systems independently.
• Write code that others can understand and reproduce.
• Learn through experimentation, debugging, and review.

There will be bugs. There will be confusing papers. There will probably be at least one implementation that works for reasons nobody completely understands.

That is part of the process.

Institution: Indian Institute of Technology Mandi

Organization: ACM Student Chapter, IIT Mandi

For questions, collaborations, or contributions, reach out to the chapter coordinators or open an issue in the relevant repository.

Learn the fundamentals. Build them yourself. Then build something new.