A structured, from-scratch journey into Deep Learning — focused on understanding, not vibe coding.
This repository documents my progression through deep learning fundamentals, starting from perceptrons and neural networks, and gradually moving toward modern deep learning architectures.
The goal of this repo is depth over speed — building strong intuition, clean implementations, and clear explanations.
At the moment, this repository focuses on building strong deep learning fundamentals through hands-on experiments and clear conceptual understanding.
Current areas of work:
Implementing neural networks from scratch and using Keras/TensorFlow
Understanding training dynamics, overfitting, and evaluation metrics
Building clean, reproducible experiments (e.g., customer churn prediction using ANN)
Strengthening intuition behind preprocessing, scaling, and model design choices
The goal is not just to build models, but to deeply understand why each step is required.
🔍 Strong fundamentals first
✍️ Explain before coding
🧠 From-scratch understanding
🚫 No vibe coding
📈 Daily incremental progress
Every concept implemented here is backed by clear reasoning and reflection.
100 Days of Deep Learning — CampusX
(Used as a structured guide, not blindly followed.)
Perceptrons & decision boundaries
Artificial Neural Networks (ANNs)
Activation functions & loss intuition
Training concepts & optimization
CNNs, RNNs, Transformers (planned)
This repository is primarily a personal learning log, but contributions, discussions, and improvements are welcome.
If you’d like to contribute:
-Fork the repository
-
Open an issue for bugs, questions, or conceptual discussions
-
Submit a pull request for improvements, corrections, or enhancements
Contributions that align with the learning-first philosophy of this repo (clear explanations, clean code, and conceptual clarity) are especially appreciated.
Browse notes/ for clean conceptual understanding
Explore notebooks/ for practical implementations
Follow the commit history to see day-by-day progress
In Progress: ANNS and CNNS have been added, RNNS coming over the weekend with a new experiment/miniproject!
Soham Mishra