I build low-latency,
failure-resilient backends.

Backend Software Engineer specializing in distributed systems. Focused on production-grade architecture, strong consistency, and sub-50ms latency targets.

Get in touchView projects
1K+ TPS
Transaction throughput
<50ms p99
Latency target
10M+ records
Pipeline scale
Kafka Outbox
Event consistency
Audit-ready
Immutable logs

Featured Projects

Real-Time Transaction Processing System

Building a payment processing system that handles high-throughput transactions with strong consistency guarantees and sub-50ms latency requirements.

JavaSpring BootPostgresKafkaDocker
Results:
  • •1K+ TPS sustained throughput
  • •Sub-50ms p99 latency
  • •Idempotency via unique keys
  • •Outbox pattern for event consistency
  • •Immutable audit ledger
Read case studyView code

Scalable Analytics & Reporting Backend

Designing a reporting system that serves complex analytical queries with aggressive caching strategies while maintaining data freshness.

JavaSpring BootPostgresRedisDocker
Results:
  • •Sub-500ms uncached queries
  • •Sub-20ms cached responses
  • •Cursor-based pagination
  • •Async background jobs
  • •Smart cache TTL/eviction
Read case studyView code

Real-Time Trust & Risk Scoring Platform

Building a real-time risk evaluation system that scores user behavior with cache-first hot paths and comprehensive failure handling.

JavaSpring BootRedisPostgresDocker
Results:
  • •Cache-first hot path design
  • •Kafka event ingestion
  • •Append-only audit logs
  • •Rate limiting + circuit breakers
  • •Bulkhead isolation
Read case studyView code

Systems Design Snapshot

Idempotency Keys

Why it exists
Prevent duplicate charges when clients retry failed requests
Failure it prevents
Double-spending, duplicate orders, inconsistent state
Tradeoff
Requires key storage and cleanup; adds latency overhead

Outbox Pattern

Why it exists
Guarantee event publishing matches database commits
Failure it prevents
Lost events, inconsistent downstream state, data drift
Tradeoff
Adds table writes; requires background polling or CDC

Cache-First Hot Path

Why it exists
Hit aggressive latency targets for high-frequency reads
Failure it prevents
Database overload, slow response times, poor UX
Tradeoff
Stale data risk; cache invalidation complexity

Let's build something.

Looking for a backend engineer who thinks about failure modes, latency budgets, and production tradeoffs.

Contact me