nairobi-os 0.3.1

Nairobi OS: Heavy Iron Data Science Infrastructure with Lagos Vision (Hardware-Accelerated Zero-Copy Rendering)

Nairobi OS: Heavy Iron AI Infrastructure

[!IMPORTANT] Linux & WSL2 Only. We do not compromise on kernel physics.

Author: Kevin Chege. Location: Nairobi License: PolyForm Noncommercial License 1.0.0

Nairobi OS is a high-performance data science operating system primitive designed for extreme resource efficiency. It enables memory-constrained environments (Edge, Containers, Serverless) to process large-scale datasets with hardware-accelerated ingestion and vectorized Rust analytics.

🚀 Version 0.3.1: The Fused Strike

This release introduces the Fused Analytics Pipeline, allowing ingestion, statistical distillation, and correlation to happen in a single, high-speed D-Bus round trip.

Key Features:

• Fused Pipeline: nairobi_os.data.pipeline() for maximum throughput.
• Zero-Copy Ingestion: Powered by memfd, io_uring, and kernel copy_file_range.
• Extreme Speed: Ingest 450MB+ datasets 12x faster than standard Pandas.
• Rayon Parallelization: Vectorized analytics leveraging multi-core hardware saturation.
• Persistent Infrastructure: Cached D-Bus connections for low-latency calls.
• Lagos Vision: Hardware-accelerated Jupyter plotting via zero-copy mmap and wgpu.

🛠️ Installation

From Wheel

pip install nairobi-os==0.3.1

Build from Source

# Clone the repository
git clone https://github.com/KevinKenya/nairobi-connector-open-source
cd nairobi-connector-open-source

# Set up Python environment
python3 -m venv .venv
source .venv/bin/activate
pip install maturin pyo3-build-config zbus anywidget traitlets

# Build the entire stack
./build_wheel.sh

# Install the forged wheel
pip install target/wheels/nairobi_os-0.3.1-py3-none-any.whl

Or install in development mode:

cd crates/nairobi-python
pip install -e .

💻 Quick Start

import nairobi_os
import json

# Ignite the refinery daemon
nairobi_os.start_refinery()

# Run a fused analytics strike
# (Ingest + Mean/StdDev/Skewness/Kurtosis + Pearson Correlation)
result_json = nairobi_os.data.pipeline(
    "data.csv",
    "target_column",
    "col1,col2" # Correlation pair
)

result = json.loads(result_json)
print(f"Mean: {result['mean']}")
print(f"Correlation: {result['pearson']}")

# Shutdown refinery
nairobi_os.stop_refinery()

📊 Performance Benchmark (v0.3.1)

Metric	Pandas (Unoptimized)	Nairobi OS	Speedup
Ingestion Latency	6.38s	0.52s	12.2x
Statistical Distillation	1.04s	0.02s	52x
Total Pipeline	6.42s	2.29s	2.8x

📦 Python API Reference

nairobi_os.start_refinery(binary_path=None, timeout=15)

Starts the Nairobi Axum Refinery daemon. Auto-discovers the binary from the bin/ directory or falls back to target/release/.

nairobi_os.stop_refinery()

Terminates the refinery daemon.

nairobi_os.connect()

Auto-configures XDG_RUNTIME_DIR, starts D-Bus (if needed on headless/Colab), and ignites the refinery.

nairobi_os.read_csv(path, delimiter=",", encoding="utf-8")

Ingests data and returns a SovereignFrame. Auto-starts refinery if offline.

nairobi_os.data.ingest(file_path)

Ingest a CSV file. Returns a handle ID (UUID string).

nairobi_os.data.crunch(handle_id, column)

Compute statistical moments on a column. Returns JSON string.

nairobi_os.data.correlate(handle_id, columns)

Compute Pearson/Spearman correlation. columns is a comma-separated string. Returns JSON string.

nairobi_os.data.pipeline(file_path, column, corr_columns)

Fused ingest + crunch + correlate in a single D-Bus round trip. Returns JSON string.

nairobi_os.data.sql_query(handle_id, query)

Execute a SQL query on an ingested dataset. Returns a new handle ID.

nairobi_os.data.free(handle_id)

Release a memfd handle.

nairobi_os.lagos.plot_inline(handle_id, width=1000, height=400)

Spawn the Lagos Vision daemon and return an interactive Jupyter widget.

⚖️ Licensing

This project is licensed under the PolyForm Noncommercial License 1.0.0. It is free for personal, educational, and research use. For commercial inquiries, please contact the author.

---

© 2026 Kevin Chege. All Rights Reserved.