FoxML Core — ML Cross-Sectional Infrastructure

License: AGPL v3 for open use. Commercial license required for proprietary or production deployments.
See LICENSING.md for details.

🔍 Reproducibility & Auditability: This system focuses on reproducibility (consistent results within expected variance) and comprehensive auditability (full tracking of inputs, configs, and outputs). Note: True bitwise determinism (identical outputs at the binary level) requires lower-level language implementations and strict control over floating-point operations.

---

⚠️ ACTIVE DEVELOPMENT — EXPECT BREAKING CHANGES
This project is under heavy active development. Breaking changes may occur without notice. APIs, configuration schemas, directory structures, and file formats may change between commits. Use at your own risk in production environments. See ROADMAP.md for current status and known issues.

🎯 Version 1.0 Definition: See FOXML_1.0_MANIFEST.md for the capability boundary that defines what constitutes FoxML 1.0.

📝 See CHANGELOG.md for recent technical and compliance changes.

---

FoxML Core is an ML infrastructure stack for cross-sectional and panel data for any machine learning applications. It provides a config-driven ML pipeline architecture designed for ML infra teams, data scientists, and researchers.

📊 Testing & Development: All testing, validation, and development work is performed using 5-minute interval data. The software supports various data intervals, but all tests, benchmarks, and development workflows use 5-minute bars as the standard reference.

Developed and maintained by Jennifer Lewis
Independent Contractor • ML Engineering • Cross-Sectional ML Systems • Systems Architecture
🔗 LinkedIn

---

Quick Overview

FoxML Core provides:

• Intelligent training pipeline with automated target ranking and feature selection
• GPU acceleration for target ranking, feature selection, and model training (LightGBM, XGBoost, CatBoost)
• Config-based usage with minimal command-line arguments
• Leakage detection system with pre-training leak detection and auto-fix
• Single Source of Truth (SST) config system - all 20 model families use config-driven hyperparameters
• Multi-model training systems with 20+ model families (GPU-accelerated)
• Reproducibility tracking with end-to-end reproducibility verification and auditability
• Local metrics tracking - Model performance metrics (ROC-AUC, R², feature importance) stored locally for reproducibility. No external data transmission, no user data collection.

For detailed capabilities: See Architecture Overview

---

Domain Focus

FoxML Core is general-purpose ML cross-sectional infrastructure for panel data and time-series workflows. The architecture provides domain-agnostic primitives with built-in safeguards (leakage detection, temporal validation, feature registry systems).

Domain Applications: Financial time series, IoT sensor data, healthcare, clickstream analytics, and any panel data with temporal structure.

For detailed domain information: See Architecture Overview

---

Intended Use

Appropriate Use Cases

• Research and experimentation (AGPL or commercial license)
• ML workflow and architecture study
• Open-source projects (AGPL)
• Internal engineering reference (commercial license)
• Production deployments (commercial license)

Not Appropriate For

• Commercial use without a license
• Unmodified production deployment without proper testing and validation

FoxML Core provides ML infrastructure and architecture, not domain-specific applications or pre-built solutions.

---

Getting Started

New users start here:

• Quick Start - Get running in 5 minutes
• Getting Started - Complete onboarding guide
• Architecture Overview - System at a glance

Complete documentation:

• Documentation Index - Full documentation navigation
• Tutorials - Step-by-step guides
• Reference Docs - Technical reference
• Technical Appendices - Deep technical topics

---

Repository Structure

FoxML_Core/
├── DATA_PROCESSING/       (Pipelines & feature engineering)
├── TRAINING/              (Model training & research workflows)
├── CONFIG/                (Configuration management system)
├── DOCS/                  (Technical documentation)
└── SCRIPTS/               (Utilities & tools)

For detailed structure: See Architecture Overview

---

Reporting Issues

For bug reports, feature requests, or technical issues:

• GitHub Issues: Open an issue (preferred for bug reports and feature requests)
• Email: jenn.lewis5789@gmail.com (for security issues, sensitive bugs, or private inquiries)

For commercial licensing or organizational engagements:
jenn.lewis5789@gmail.com

---