The Arclight Project Roadmap

Note

Arclight follows principles of MVP development and separation of concerns. The main Arclight model will be the initial release, developed, optimized, and made available in a standalone ONNX format. A quantized version will then be developed as an integrated Streamlit app packaged in a deployable Docket container. The fully integrated Flutter application will be the final release, initially targeting Android, Windows, and Linux systems.

Phase 1: Model Development & Optimization (Google Colab)

Train, validate, and optimize machine learning models within cloud notebook environments before exporting for on-device execution, publishing source code under an open-source license.

Tasks:

• Colab Model Development: Develop, train, and validate initial model inside Google Colab.

• Optimization & Quantization: Apply post-training quantization and model compression techniques to minimize memory footprint and optimize execution speed for target end-user devices.

• ONNX Export: Export tuned models into the industry standard ONNX format required for widespread mobile and desktop release.

• Open-Source Licensing: Host model source code and assets on GitHub under the GNU General Public License v3.0 (GPL v3.0).

Phase 2: UI/UX Proof of Concept (Streamlit & Docker)

Prototype the user interface and validate operational workflows quickly through a web-based Proof of Concept (POC).

Tasks:

• Streamlit Prototype: Build a functional Streamlit interface to mock up layouts, data ingestion workflows, and visualization formats required for different end-use cases.

• Workflow Validation: Test end-to-end data processing, user interaction flows, and structured decision-support outputs within the Streamlit workspace.

• Docker Containerization: Package the Streamlit application into a Docker container for isolated testing, streamlined sharing, and collecting feedback from epidemiologists and stakeholders.

Phase 3: Flutter Application Development & Integration

Construct the production cross-platform native application framework and integrate local machine learning models into the Dart pipeline.

Tasks:

• App Scaffolding & Front-End: Develop responsive front-end UI components in Flutter supporting Android, Linux, and Windows targeting cross-platform consistency.

• Dart Data Pipeline: Implement data import, sanitization, and preprocessing functions in Dart to prepare user inputs for model consumption.

• On-Device Model Integration: Integrate exported .onnx models into the Flutter application for privacy-preserving, local execution.

• Secure Local Storage: Implement encrypted local databases (e.g., SQLite or Hive) with local encrypted backups to external storage media like USB drives.

Phase 4: Integration Testing & Multi-Platform Deployment

Conduct rigorous system-wide testing and publish production application builds across supported platforms and repositories.

Tasks:

• Integration Testing: Perform comprehensive integration testing across Android, iOS, and Windows, verifying on-device model accuracy, UI responsiveness, PHI censoring algorithms during PDF/CSV exports, and database security.

• App Store & Open-Source Distribution: Publish finalized production binaries to app stores (Google Play Store, Apple App Store, Microsoft Store) and release full source code and build distributions on GitHub.