from-scratch-implementation

Companion code for Machine Learning From Scratch — 10 core ML algorithms built from scratch with NumPy, compared with Scikit-learn and PyTorch.

An OS which is all about learning!

A from-scratch AlphaFold2 in PyTorch designed to make one of the most important and complex ML architectures readable, hackable, and ablatable.

Educational, from-scratch implementation of a LLaMA-style LLM using PyTorch to explore Transformer architecture fundamentals.

A from-scratch implementation of a feedforward neural network in C# (.NET 8) without using any machine learning frameworks.

A convolutional neural network (CNN) built from scratch using only NumPy to classify handwritten digits from the MNIST dataset.

From-scratch C++17 LiDAR + camera perception — sector RANSAC, LiDAR-inertial SLAM, MOT, probabilistic traversability, CBF safety. 13 engineering deep-dives.

"Learn Linear Regression: A Python implementation from scratch with dataset generation and visualization" as it's both informative and engaging.

Kademlia POC Implementation

A minimalist, CUDA-native deep learning library featuring a custom autograd system and a PyTorch-inspired API, built to explore framework internals and GPU-accelerated computing.

A decoder-only Transformer built from scratch using CuPy only — no PyTorch, no autograd, no magic. Every forward pass, backward pass, and gradient derived and implemented by hand. Includes full training loop, Adam optimizer, LayerNorm, and causal self-attention on GPU.

From-scratch PyTorch implementations of transformer components, BPE, decoding strategies, manual backprop, and RL.

ML algorithms implemented from scratch in Python, with small projects for better understanding.

This project demonstrates how to build and train a feedforward neural network from scratch using only NumPy, without any high-level deep learning libraries like TensorFlow or PyTorch. The model is trained on the MNIST digit classification dataset and achieves competitive accuracy.

Manual implementation of backpropagation on a custom computational graph with gradient checking. Benchmarks Vanilla SGD, Momentum, and Adam optimizers from first principles using NumPy.

Implementation of KNN and Gaussian Naive-Bayes algorithms to classify phishing URLs. Built from scratch and compared with scikit-learn versions.

A from-scratch implementation of a high-performance, decoder-only Transformer language model, built on top of the custom Phoenix Engine framework.

A curl implementation written in python3.

MapReduce POC Implementation

Predicting clinical trial duration from registration data. Neural network built from scratch in NumPy, benchmarked against sklearn and Keras.