v0.3.0

Architecture

FacilFlow system architecture and component design.

Platform Context

graph TB
    subgraph "Facilis Platform"
        Jarvis["Jarvis AI
jarvis.facilis.ai"]
        FacilFlow["FacilFlow
flow.facilis.ai"]
        Firebase["Firebase
Auth & Firestore"]
    end

    subgraph "ML Platform"
        MLflow["MLflow
Model Registry"]
    end

    subgraph "Edge Layer"
        FacilEdge["FacilEdge Agents"]
    end

    subgraph "Plant Floor"
        PLC["PLCs"]
        SCADA["SCADA"]
        Historians["Historians"]
    end

    Jarvis <-->|REST API| FacilFlow
    Jarvis --> Firebase
    FacilFlow --> Firebase
    FacilFlow <-->|gRPC/MQTT| FacilEdge
    FacilFlow --> MLflow
    FacilEdge --> PLC
    FacilEdge --> SCADA
    FacilEdge --> Historians

    style Jarvis fill:#0D9488,color:#fff
    style FacilFlow fill:#22C55E,color:#fff
    style FacilEdge fill:#F59E0B,color:#fff
    style MLflow fill:#EC4899,color:#fff

System Components

Frontend Pages (28 total)

Section Page Purpose
Auth LoginPage Firebase email/password login
SignupPage New user registration
ProfilePage User profile management
SettingsPage App and account settings
Connect ConnectorsPage Manage data connectors
ConnectorHubPage Browse available connector types
AirbyteCatalogPage PyAirbyte connector catalog
TelegrafPluginsPage Telegraf input plugin browser
APIRepositoryPage Custom API connector repository
AgentsPage Edge agent list and status
AgentDetailPage Single agent details and metrics
AgentConfigPage Agent configuration editor
DataSourcesPage Registered data source browser
Infrastructure ClustersPage Compute cluster management
Build PipelinesPage Pipeline list, create, manage
MLModelsPage ML model registry and lifecycle
SchemaRegistryPage Data schema definitions
Analyze DataExplorerPage SQL/InfluxQL query interface
DashboardPage Custom dashboard builder
DataQualityPage Data quality rules and scores
LineageExplorerPage Data lineage graph visualization
Operate AlertsPage Alert rules and notifications
SchedulerPage Pipeline scheduling (cron)
StoragePage Storage provider management
ObjectStoreManagerPage Object/file browser for buckets
AuditLogsPage Activity audit trail
GovernancePage Data governance policies

Backend Services

Service Purpose
PipelineService Pipeline CRUD, execution, status tracking
AgentService Edge agent registration, heartbeat, commands
ClaudeService AI assistant via Claude Code CLI spawn with OAuth
ChatService Conversation history and context
DataSourceService Connector and data source management
StorageService Multi-cloud object storage operations
MLService Model training, inference, approvals, feature store
SchedulerService Cron-based pipeline scheduling
AlertService Alert rule evaluation and notification dispatch
DashboardService Dashboard CRUD and widget data
QualityService Data quality rule execution and scoring
LineageService Data lineage capture and graph queries
SchemaService Schema registry and validation
AuditService Activity logging and audit trail
ClusterService Compute cluster provisioning and monitoring

Data Layer

Store Purpose
PostgreSQL Pipelines, agents, users, configs, audit logs (Drizzle ORM)
InfluxDB 3 Time-series metrics and telemetry (Arrow Flight SQL)
MinIO/S3/Azure/GCS Object storage for artifacts, models, files
Redis Cache, pub/sub, real-time state, job queues

MCP Tools

Tool Set Tools Purpose
pipeline_tools list, create, start, stop, delete Pipeline management via Jarvis
query_tools explore_catalog, query_data Data exploration via Jarvis

Data Flow

flowchart LR
    PLC[Plant Floor] --> Edge[FacilEdge]
    Edge -->|gRPC| Receiver[Data Receiver]
    Receiver --> Validate[Schema Validator]
    Validate --> Store[(InfluxDB)]
    Validate --> Quality[Quality Checks]
    Validate --> Lineage[Lineage Capture]
    Store --> API[Query API]
    API --> UI[Web UI]
    API --> Jarvis[Jarvis AI]
    Store --> ML[ML Pipeline]
    ML --> MLflow[MLflow Registry]
    MLflow --> Inference[Inference Service]

ML Pipeline Flow

flowchart LR
    Data[(InfluxDB / PostgreSQL)] --> Features[Feature Engineering]
    Features --> Train[ML Training]
    Train --> MLflow[MLflow Registry]
    MLflow --> Approve[Approval Gate]
    Approve --> Deploy[Inference Service]
    Deploy --> Results[Predictions]
    Results --> Store[(Storage)]

Real-time Architecture

Socket.IO namespaces and events:

Namespace Events Purpose
/pipelines pipeline:status, pipeline:metrics Pipeline execution updates
/agents agent:heartbeat, agent:status Edge agent health
/dashboards dashboard:data, widget:refresh Dashboard live data
/alerts alert:triggered, alert:resolved Alert notifications
/clusters cluster:status, cluster:metrics Cluster health
esc