Backend Engineer
Building reliable integrations and observable distributed systems.
Backend engineer with around 9 years of experience across Java/Spring and TypeScript/Node.js/NestJS backend systems that connect multiple services and operate under real-world constraints. Focused on system behavior, data flow, failure modes, transaction boundaries, and operational constraints rather than treating any one technology stack as the goal.
Focus
Engineering focus
- Backend Engineering
- Distributed & Event-driven Systems
- Enterprise Integration
- Production Troubleshooting
- Observability
- Kubernetes Platform Operations
Projects
Featured projects
Written as problem, architecture, decisions, failure handling, and result — not a technology list.
RelayHub
LiveAn event-integration platform: events are ingested, validated, transformed, and delivered to downstream targets with retry, dead-letter handling, replay, and metrics — the same behavior this site's Live Lab lets you trigger yourself.
- Java
- Spring Boot
- Kafka
- PostgreSQL
English Core Speaking
LiveA full-stack English speaking practice service: a NestJS backend and Vue 3 frontend, with OAuth-based sign-in and an external AI/LLM API integration for evaluating spoken responses.
- NestJS
- Vue 3
- PostgreSQL
- Prisma
Live Lab
RelayHub live engineering lab
Generate a synthetic event and watch it move through ingestion, validation, delivery, retry, and DLQ/replay.
RelayHub pipeline snapshot
Synthetic demo data- Requests / sec
- 24
- Success rate
- 98.2%
- P95 latency
- 214 ms
- Retries
- 2
- DLQ
- 1
Case Studies
Selected case studies
Abstracted production engineering scenarios: investigation, root cause, trade-offs, and lessons learned.
A Race Condition in Distributed Delivery Processing
An asynchronous delivery pipeline occasionally left a downstream record in an inconsistent state when a delete and a create for the same logical entity raced each other. This is a composite, abstracted scenario representative of enterprise integration work — no real company, customer, or system names are used.
Timeout and Concurrency Issues in an External-Approval Rollback Flow
A workflow that depended on an external system's approval step had a rollback path that assumed the approval call would complete quickly and exactly once. Neither assumption held under real conditions. This is a composite, abstracted scenario — no real company, customer, or system names are used.
Platform
Architecture & platform snapshot
Every project here runs on the same Hetzner k3s cluster, behind a single shared Gateway, deployed by its own CI pipeline.
More