·4 min read·Shipped
Diagnosing a Production Freeze: Connection Pool Starvation Across Service Boundaries
A one-hour production freeze, traced back to a transaction that outlived its own database connection.
I'm a Backend Software Engineer at Texas Instruments focusing on transaction correctness, database & ORM performance engineering, and distributed systems.
Selected Work & Impact
95%
ORM & Batching Latency Cut
·PerformanceReduced 45+ min workflows to ~2 min at Texas Instruments via JDBC batching, Persistable<T>, and correlated SQL updates.
10K+
11-State Modular Monolith
·Production / ArchitectureArchitected queue-less lifecycle orchestrator automating sales workflows across 10,000+ participants, saving ~$120K/yr.
Raft
Distributed Consensus in Go
·Distributed SystemsFrom-scratch Raft implementation in Go, currently focused on leader election, randomized election timeouts, heartbeats, and the path toward log replication.
~30s
Metadata Knowledge Graph
·Graph & DBIndexed 16,000+ artifacts and 30,000+ dependency edges with PostgreSQL GIN reverse search, slashing dependency analysis from months to seconds.
Technical Writing & Investigations
4 articles·3 min read·Shipped
What Persistable<T> Taught Me About Hibernate's Entity State Detection
Why Hibernate issued a SELECT before every INSERT on a composite-key entity, and what fixing it unlocked.
·3 min read·Note
Designing Data Models to Minimize LLM Inference Cost
Some early lessons from structuring data so an AI-assisted product calls the model less often, not just cheaper.
·3 min read·In progress
Implementing Raft From the Paper: Notes on Leader Election
Starting a from-scratch Raft implementation to actually understand consensus, not just read about it. First stop: leader election.
Currently Building & Exploring
Raft in Go
Leader election → log replication → persistence
Distributed Systems
Consistency, Sharding, MapReduce