kiteml-ai 1.0.1

Train production-grade ML models with a single line of code — intelligent AutoML for everyone.

🪁 KiteML

Train production-grade ML models with a single line of code.

---

KiteML is an intelligent AutoML framework that automates the entire ML pipeline — from raw data to production-ready models. It handles preprocessing, feature engineering, model selection, training, evaluation, serving, and deployment, all through a clean Python API and powerful CLI.

---

Features

Category	Capabilities
Core ML	Auto preprocessing, model selection, cross-validated training, evaluation reports
Intelligence	Explainability (SHAP/feature importance), imbalance detection, data profiling
Production	FastAPI serving, ONNX export, Docker packaging, batch & real-time inference
CLI	14 subcommands — train, serve, predict, profile, doctor, and more
Integrations	WandB, MLflow, plugin SDK for custom extensions
Governance	Model cards, audit logging, experiment tracking
I/O Formats	CSV, Excel, JSON, Parquet

---

Requirements

• Python 3.11+ (3.11, 3.12, and 3.13 are officially supported)

---

Installation

pip install kiteml-ai

Extras

pip install kiteml-ai[serving]   # FastAPI model server
pip install kiteml-ai[onnx]      # ONNX export support
pip install kiteml-ai[wandb]     # Weights & Biases tracking
pip install kiteml-ai[mlflow]    # MLflow experiment tracking
pip install kiteml-ai[all]       # Everything

Note: The PyPI package is kiteml-ai. The Python import remains import kiteml — same as scikit-learn → sklearn.

---

Quick Start

Python API

from kiteml import train

# Classification
result = train("data.csv", target="label")
print(result.summary())
result.save_model("my_model.pkl")

# Regression
result = train("housing.csv", target="price", problem_type="regression")
print(result.summary())

# Make predictions
predictions = result.predict(new_data)

CLI

# Train a model
kiteml train data.csv --target label

# Train with options
kiteml train data.csv --target price --type regression --save model.pkl

# Serve a model
kiteml serve model.pkl --port 8000

# Profile your dataset
kiteml profile data.csv

# Run diagnostics
kiteml doctor

---

Architecture

kiteml/
├── core.py              # Main train() function
├── preprocessing/       # Auto cleaning, encoding, scaling
├── models/              # Model selection & training
├── evaluation/          # Metrics & reporting
├── intelligence/        # Explainability, profiling, imbalance detection
├── serving/             # FastAPI production server
├── deployment/          # ONNX, Docker, packaging
├── monitoring/          # Drift detection & performance tracking
├── experiments/         # Experiment tracking & logging
├── plugins/             # Extensible plugin SDK
├── governance/          # Model cards & audit logging
└── cli/                 # 14-command CLI ecosystem

---

Documentation

Full documentation is available at https://kiteml.github.io/kiteml.

• Getting Started
• Usage Guide
• CLI Reference
• API Reference

---

Contributing

We welcome contributions! See CONTRIBUTING.md for guidelines.

# Development setup
git clone https://github.com/Priyatham27/kiteML.git
cd kiteml
pip install -e ".[dev]"
pytest tests/

---

License

KiteML is released under the MIT License.

---

_{Built with care by the KiteML Team}