Services

The Ona Platform is built on a distributed architecture of modular agents, each specializing in a particular task within the Ona Intelligence Layer. These services work together to transform raw operational data into actionable business intelligence through a coordinated, event-driven pipeline.

Each service operates as an independent, specialized agent that:

• Focuses on a single responsibility: From data collection to ML model training, each agent excels at its specific domain
• Communicates through events: Services interact via events, API calls, and message queues, enabling loose coupling and scalability
• Maintains state independently: Each agent manages its own data, configuration, and operational state
• Scales autonomously: Services can scale independently based on their specific workload patterns
• Contributes to the intelligence layer: Together, these agents form the Ona Intelligence Layer that powers the OODA (Observe, Orient, Decide, Act) workflow

This modular architecture enables the platform to handle complex energy management tasks—from real-time anomaly detection to 30+ day predictive forecasting—by orchestrating specialized agents that each bring deep expertise to their domain. Whether collecting data from solar inverters, standardizing multi-OEM formats, training customer-tailored ML models, or generating automated work orders, each service agent plays a critical role in delivering intelligent energy management capabilities.

Data Collection Services

Data Ingestion Service

AI agent that serves as the secure entry point for incoming data, providing pre-signed URLs for direct uploads.

Key Features:

• Secure entry point for data uploads
• Pre-signed URL generation for secure uploads
• Structured logging and metrics

Documentation: See services/dataIngestion/README.md in the platform repository.

Huawei Historical Service

AI agent that collects historical data from Huawei FusionSolar inverters, transforms it to ONA platform format, and uploads to storage.

Key Features:

• Huawei FusionSolar API integration
• Device discovery and data collection
• Weather data enrichment
• Direct upload to historical data storage

Documentation: See services/huaweiHistorical/README.md and services/huaweiHistorical/HUAWEI_API_DOCS.md in the platform repository.

Weather Data Updater Service

AI agent that automatically updates weather data for South African cities by fetching missing data from Visual Crossing API.

Key Features:

• Dynamic date range calculation
• Cost-optimized API usage
• Duplicate prevention
• Support for Cape Town, Johannesburg, and Durban

Documentation: See services/weatherDataUpdater/README.md in the platform repository.

Data Processing Services

Data Standardization Service

AI agent that processes files from historical storage, detects OEM type, standardizes schema, and saves to standardized storage.

Key Features:

• OEM type detection (Huawei, Enphase, Solarman, Telkom/Huawei)
• Schema normalization
• Error code mapping to categorical types
• Timestamp aggregation

Documentation: See services/dataStandardizationService/README.md and services/DATA_STANDARDIZATION_SERVICE_PLAN.md in the platform repository.

Interpolation Service

AI agent that performs ML-based solar data interpolation with comprehensive gap analysis and configuration-driven architecture.

Key Features:

• Configuration-driven interpolation architecture
• Gap analysis and pattern detection
• Multiple interpolation methods (Spline, Gaussian Process, Physics-based, Multi-output regression)
• Adaptive interpolation with R² > 0.99
• Weather integration and solar physics constraints

Documentation: See services/interpolationService/README.md in the platform repository.

Weather Cache Service

AI agent that fetches weather data for all active locations and caches results.

Key Features:

• Asynchronous concurrent API calls
• Rate limiting for external APIs
• Centralized cache storage
• Database integration for location management

Documentation: See services/weatherCache/README.md in the platform repository.

Machine Learning Services

Global Training Service

AI agent that trains forecasting models using customer validation optimization through a hybrid architecture.

Key Features:

• Site-level and device-level training
• Customer validation optimization strategy
• GPU-accelerated model training
• Automatic customer discovery and quality filtering
• Model registry for forecasting API

Documentation: See services/globalTrainingService/README.md in the platform repository.

Forecasting API Service

AI agent that provides solar energy forecasting via API endpoint, loading ML models and generating predictions.

Key Features:

• TensorFlow and Prophet integration
• Redis caching for forecast results
• 30+ day forecasting capabilities
• API specification with OpenAPI

Documentation: See services/forecastingApi/README.md in the platform repository.

Application Services

Terminal API Service

AI agent that serves as the central API handler for the entire O&M OODA workflow, routing requests to appropriate sub-handlers.

Key Features:

• Unified entry point for OODA workflow
• Asset management (Create, Read, List)
• Fault detection and AI diagnostics
• Maintenance scheduling and work orders
• Database integration for state management

Documentation: See services/terminalApi/README.md in the platform repository.

Edge Device Registry Service

AI agent that manages device discovery, registration, and capability detection for distributed edge devices.

Key Features:

• Device discovery and registration
• Capability detection
• Device status tracking
• RESTful API interface

Documentation: See services/edge-device-registry/app.py in the platform repository.

Energy Analyst RAG Service

AI agent that provides RAG-powered energy policy and regulatory compliance analysis.

Key Features:

• Regulatory compliance queries
• Policy analysis
• Document retrieval and analysis
• AI-powered insights

Documentation: See services/energyAnalystRag/README.md and related deployment guides in the platform repository.

Service Architecture

All services follow a consistent architecture pattern:

• Container-based: Built as Docker containers
• Serverless: Deployed as serverless functions
• Base Image: Most services inherit from ona-base image
• Logging: Structured logging and metrics
• Metrics: Metrics integration and monitoring
• Storage Integration: Standardized storage structure
• Event-Driven: Events trigger downstream processing

Service Dependencies

Base Image (ona-base)

Common dependencies and utilities shared across services:

• Cloud service SDKs for infrastructure integration
• pandas for data processing
• Structured logging and metrics tools
• Common utilities in utils/common.py

Documentation: See services/base/README.md in the platform repository.

Service Integration Flow

Data Collection → Data Standardization → Training → Forecasting
     ↓                    ↓                  ↓            ↓
huaweiHistorical → dataStandardization → globalTraining → forecastingApi
weatherDataUpdater → Service            → Service      → Service
     ↓
weatherCache
     ↓
interpolationService
     ↓
terminalApi (OODA Workflow)

Getting Started

For detailed documentation on each service, refer to the README files in /home/shingai/platform/services/:

1. Navigate to the service directory
2. Review the README.md file
3. Check environment variables and configuration
4. Review build and deployment instructions

Support

For questions or issues with services:

• Email: support@asoba.co
• GitHub Issues: https://github.com/AsobaCloud/platform/issues
• Documentation: https://docs.asoba.co

© 2025 Asoba Corporation. All rights reserved.