Leading Solana Integrations Infrastructure for Automated Accounting

Integral was running into real issues with its Solana integration and the underlying temporal queue infrastructure. Most of the existing system had been built with EVM chains in mind, using scaling patterns that work well there but break down quickly on Solana.

As transaction volume increased, customers were seeing:

• •Missed and unclassified journal entries
• •Incomplete or incorrect Solana accounting data
• •Customer synchronizations taking 7–10 days
• •Growing strain on a polling-heavy RPC model

On top of that, Integral had active partnerships with major players in the Solana ecosystem. The Solana integration wasn't just "nice to have", it needed to be correct, fast, maintainable, and demo-ready for enterprise partners.

I was brought in to fix it.

I took ownership of the Solana integration end-to-end and rebuilt the critical pieces properly, starting from how Solana actually works at the protocol level.

On the Solana side, this included:

• •Correctly handling associated token accounts
• •Classifying staking instructions
• •Breaking down multi-instruction transactions into proper individual journal entries
• •Fixing unclassified and missed transactions at the accounting layer

On the infrastructure side, I focused heavily on performance and scale:

• •Reduced RPC usage significantly by moving from polling to gRPC streaming once historical data was archived
• •Eliminated the need to re-run full synchronizations
• •Designed and implemented a cyclic job queue that batches work across multiple organizations
• •Reprioritized accounts dynamically based on transaction volume and other metrics
• •Implemented transaction rollups to reduce redundant processing
• •Migrated key workflows from Bull to Temporal, setting the groundwork for what became Integral's v2 system

The goal wasn't just to make Solana "work," but to make it boring: predictable, fast, and something the rest of the platform could rely on.