A curated, developer-focused knowledge base for building an in-house blockchain data platform. Use this README to navigate core concepts, PRDs, architectures, specifications, and implementation epics.

• New here? Start with Concepts to ground terminology and mental models.
• Evaluating feasibility? Read the PRDs, then the Architecture docs.
• Building a PoC? Jump straight to Processing Epics (Epic-001).

• docs/concepts.md — Developer guide to chains, wallets, DEXs, pools, bridges.
• docs/wallet_data_specs.md — Comprehensive wallet data types, sources, and implementation patterns.

• docs/prd_data_platform.md — Full-control wallet data platform PRD (end-to-end vision).
• docs/processing_layer_prd.md — Processing layer PRD (decode, normalize, enrich, integrity).

• docs/ingestion_layer_architecture.md — Ingestion architecture (nodes, listeners, queues, bronze → gold).
• docs/processing_layer_architecture.md — Processing layer architecture with tech stack rationale.

• docs/ingestion_layer_spec.md — Decision-oriented ingestion spec: run vs lease, fetch patterns, SLOs.
• docs/event_data_spec.md — Event speed/scale proofs, storage math, and implementation patterns.

• docs/processing_epics/README.md — Epic index and phase plan.
  • EPIC-001: PoC Foundation — Free-tier PoC from scratch; end-to-end demo.
  • EPIC-002: Core Decoding — ABI registry, batch decoding, normalization, retries.
  • EPIC-003: Data Enrichment — Token/NFT metadata, pricing, protocol labeling.
  • EPIC-004: DeFi Analytics — Swaps, approvals, NFT activity, LP analytics.
  • EPIC-005: Reorg Management — Canonical flags, rollback/recovery, monitoring.

• Prototype: Follow EPIC-001 step-by-step to get a working PoC with free resources.
• Scale: Read ingestion/processing architecture and specs; progress through Epics 002–005.
• Decide: Use decision sections (SLOs, trade-offs, CAP notes) to justify vendor/infra choices.

• Consistency-first system of record (Postgres) with CP trade-offs; AP-friendly analytics (ClickHouse, Parquet).
• Layered storage: Bronze (object store/Parquet) → Silver/Gold (ClickHouse) with canonical flips.
• Idempotency and reorg safety on every row; versioned ABIs and protocol registries.
• Start simple; optimize with Kafka/Redpanda, Rust/Go workers as throughput grows.

• Event throughput and storage sizing math: docs/event_data_spec.md
• Ingestion run-vs-lease decisions: docs/ingestion_layer_spec.md
• Processing tech stack rationale: docs/processing_layer_architecture.md

This repository is a knowledge base for internal and client-facing proposals and PoCs. Add a license here if distributing externally.