amp.md

flowchart LR
    %% Data Sources
    Node["EVM Node (Local or RPC)"]

    %% AMP Core
    subgraph AMP["AMP Indexer"]
        Extract["Block and Log Extraction (Parallel)"]
        Raw["Raw EVM Datasets Parquet in Object Storage"]
        Compact["Compaction Optimized Parquet"]
    end

    %% Transformations
    subgraph Transform["Derived Data Layer"]
        Views["Materialized Views SQL Transforms"]
        UDFs["Custom Logic TypeScript UDFs"]
    end

    %% Serving Layer
    subgraph Serving["Application and Analytics"]
        DB["App Databases Postgres or D1"]
        API["APIs and Indexers"]
        BI["Analytics SQL Playground"]
        Explorer["Block Explorer Live UI"]
    end

    %% Cross-chain
    Cross["Cross-Chain Joins Multi-Dataset SQL"]

    %% Flow
    Node --> Extract
    Extract --> Raw
    Raw --> Compact
    Compact --> Views
    UDFs --> Views
    Views --> DB
    Views --> BI
    DB --> API
    API --> Explorer
    Views --> Cross

Loading

1. Raw datasets

• Canonical EVM data as Parquet

2. Derived datasets

• Materialized views built via SQL transforms

3. Continuously updated physical tables

• Application databases

4. Stream processed data into Postgres, D1, etc.

• Cross-chain analytics • Join datasets across multiple chains at the SQL level