Blockchain Indexing System

Multi-Chain Treasury Management

Mensari’s treasury platform needed to track assets across Ethereum, Polygon, and 8+ additional blockchain networks in real time. The original architecture relied on batch polling — periodically querying each chain’s RPC endpoints for new transactions — which introduced latency, missed transactions during high-throughput periods, and scaled linearly with the number of supported chains. As the platform expanded to new networks, each integration took approximately three weeks of engineering effort to build, test, and stabilize.

Event-Driven Architecture

I redesigned the architecture, migrating from batch polling to an event-driven model centered on Alchemy webhooks, which transformed the system’s throughput and reliability. Instead of pulling data on a schedule, the indexer receives push notifications for on-chain events and processes them through a pipeline backed by Redis for deduplication and ordering. This architecture achieved a 10x throughput improvement while reducing infrastructure costs, since the system only does work when there are actual transactions to process rather than continuously polling idle chains.

Chain-Agnostic Data Layer

The most significant architectural decision was building a chain-agnostic abstraction layer that normalized transaction data from different blockchain networks into a unified schema. Each chain adapter handles network-specific encoding, confirmation thresholds, and event formats, while the core indexing pipeline operates on a single canonical data model. This abstraction compressed new chain integrations from three weeks down to three days. Wallet notifications now arrive in sub-second latency, giving treasury managers real-time visibility into cross-chain fund movements.

Multi-chain systems only become manageable when you stop treating each chain as special and start treating chain-specific behavior as a pluggable detail behind a stable abstraction.