A high-octane, precision-based arcade racing experience built on drift physics, split-second control, and dodging fatal obstacles.

DeadDrift is an adrenaline-fueled racing game where survival requires perfect traction control and master-class drifting. Slide through tight corners, outrun chaotic elements, and maintain momentum to avoid burning out. Every turn is a gamble between a record-breaking speed boost or a total crash.

• Advanced Drift Physics: Immersive steering mechanics designed for tight cornering and satisfying momentum management.
• Deadly Environments: Dynamic tracks filled with hazardous turns, obstacles, and high-stakes racing lanes.
• Optimized Performance: Smooth rendering and highly responsive inputs to catch every split-second slide.

A powerful, single portable binary that bundles an interactive developer interface, local socket inspection, and an embedded offline LLM.

• Zero Cloud Dependencies: Fully functional local LLM execution with no external API keys or internet required.
• Portable Pipeline: Compiled entirely into a single standalone binary for absolute zero-configuration setups.
• Socket Inspector: Full-featured interactive interface to analyze and debug local network traffic seamlessly.

A custom-built, lightweight artificial neural network architecture constructed from scratch to model deep learning nodes, mathematical weights, and activation functions.

This implementation features an independent Artificial Neural Network(ANN). By stripping away heavy, abstract frameworks, this project isolates the fundamental mechanics of machine learning to demonstrate exactly how data propagates, adjusts, and learns across hidden layers.

• Custom Backpropagation: Algorithmic calculation of partial derivatives to compute error gradients and dynamically fine-tune network weights.
• Matrix Operations: Highly optimized mathematical structures handling dot products and forward feed processing with minimal overhead.
• Activation Nodes: Integrated mathematical activation routines designed to process complex non-linear classification boundaries.