diff --git a/CONTRIBUTING.md b/CONTRIBUTING.md
new file mode 100644
index 000000000..b218ae654
--- /dev/null
+++ b/CONTRIBUTING.md
@@ -0,0 +1,89 @@
+# Contributing to BitNet.cpp
+
+Thank you for contributing to BitNet.cpp, Microsoft's 1-bit LLM inference framework running on CPU.
+
+## Project Overview
+
+BitNet.cpp provides hardware-accelerated inference for 1-bit LLMs (like BitNet-b1.58-2B-4T) on CPU via llama.cpp compatibility. It supports x86_64 (AVX2/AVX512) and ARM64 (NEON) architectures.
+
+## Development Setup
+
+### Prerequisites
+
+- C++17 compatible compiler (GCC 9+, Clang 12+, MSVC 2019+)
+- CMake 3.15+
+- Python 3.10+ (for model conversion)
+
+### Building
+
+```bash
+git clone https://github.com/Scottcjn/BitNet.git
+cd BitNet
+mkdir build && cd build
+cmake ..
+make -j$(nproc)
+
+# Or use the build script
+./bitnet.sh build
+```
+
+### Downloading Models
+
+```bash
+# Download a 1-bit model from Hugging Face
+python download-model.py microsoft/BitNet-b1.58-2B-4T
+```
+
+### Running
+
+```bash
+# Interactive mode
+./bitnet -m models/bitnet-b1.58-2b-4t.gguf -p "Your prompt here"
+
+# Chat mode
+./bitnet -m models/bitnet-b1.58-2b-4t.gguf --color -ins -r "User:"
+```
+
+## Code Style
+
+- C++17 standard
+- Follow LLVM/Google style guide
+- Use `uint64_t`/`size_t` for bit operations
+- SIMD intrinsics in separate compilation units
+- Header-only utilities allowed
+
+## Testing
+
+```bash
+# Build tests
+cmake .. -DBITNET_BUILD_TESTS=ON
+make -j$(nproc)
+
+# Run tests
+ctest --output-on-failure
+
+# Benchmark
+./benchmark --model models/bitnet-b1.58-2b-4t.gguf --n-prompt 512 --n-gen 128
+```
+
+## Architecture Notes
+
+- llama.cpp compatible model format (GGUF)
+- Weight-only quantization (WOQ) for 1.58-bit models
+- SIMD kernels: AVX2 (x86), NEON (ARM64)
+- KV cache and prompt processing fully supported
+
+## Submitting Changes
+
+1. Fork the repository
+2. Create a branch: `git checkout -b fix/your-fix`
+3. Ensure code compiles and tests pass
+4. Submit a pull request
+
+## Ideas for Contributions
+
+- Additional architecture support (POWER8/POWER9 via VSX)
+- More SIMD kernel optimizations
+- New quantization formats
+- WASM/WebGPU backend
+- Memory footprint improvements