Specialized in backend systems engineering, distributed architectures, and performance-critical applications Proven track record architecting scalable microservices, Building Systems That Stay Correct Under Load, and optimizing production systems
I'm a Backend Systems Engineer specializing in distributed architectures, event-driven systems, and performance-critical applications.
I don't build CRUD apps. I build systems that behave correctly under load, failure, and concurrency.
My focus areas:
- Distributed architectures — microservices, service meshes, data consistency
- Event-driven systems — Kafka-based async pipelines, exactly-once semantics
- Concurrency & orchestration — DAG execution, parallel task scheduling
- Production-grade reliability — circuit breakers, retries, DLQ, self-healing
Ranked by engineering complexity and system depth
A distributed orchestration engine that models how real-world trading systems safely execute orders under strict latency and correctness constraints.
⚠️ This is not a trading UI — it is a control plane ensuring safe execution under concurrency.
The Core Problem: A trade is not a single operation. Before execution, it must pass risk validation, margin checks, and compliance rules — all in parallel, within strict latency bounds, and with consistency under failure.
[ Order Received ]
│
▼
┌─────────────────────────────────────────────┐
│ Pre-Trade (Sync, gRPC) │
│ Risk Check ──┬── Margin Check ──┬── KYC │
│ └──── Compliance ──┘ │
└──────────────────────┬──────────────────────┘
│ All pass?
▼
┌─────────────────────────────────────────────┐
│ Orchestration Layer (DAG Engine) │
│ CompletableFuture.allOf() + Retry/CB │
└──────────────────────┬──────────────────────┘
│
▼
┌─────────────────────────────────────────────┐
│ In-Memory Matching Engine │
│ Price-time priority + Partial fills │
└──────────────────────┬──────────────────────┘
│
▼
┌─────────────────────────────────────────────┐
│ Post-Trade (Async, Kafka) │
│ Ledger ─── Notifications ─── Analytics │
└─────────────────────────────────────────────┘
| Engineering Concern | Implementation |
|---|---|
| Parallel Validation | gRPC + CompletableFuture.allOf() |
| Failure Resilience | Retries + Circuit Breakers + DLQ |
| Exactly-Once | Atomic DB + Kafka offset coordination |
| State Machine | Explicit order lifecycle transitions |
| Async Decoupling | Kafka-based post-trade pipeline |
Stack: Java Spring Boot Apache Kafka gRPC Redis PostgreSQL
Workflows are not hardcoded pipelines — they are defined as DAGs and executed dynamically across a distributed worker pool.
[ Workflow Definition (DAG) ]
│
┌─────▼─────┐
│ Scheduler │ ← Topological dependency resolution
└─────┬─────┘
│
┌──────▼──────┐
│ Kafka Bus │ ← Decouples orchestration from execution
└──────┬──────┘
│
┌────────▼────────┐
│ Worker Pool │ ← Horizontally scalable via K8s HPA
│ (Consumer Lag) │
└────────┬────────┘
│ Task fails?
┌──────▼──────┐
│ Retry → │
│ DLQ → │ ← Self-healing fallback chain
│ Alert │
└─────────────┘
| Engineering Depth | Detail |
|---|---|
| DAG Scheduling | Topological sort → dependency-aware execution |
| Event-Driven Core | Kafka decouples orchestration from execution |
| Exactly-Once | Atomic DB writes + Kafka offset commit |
| Self-Healing | Failed task detection → retry → DLQ fallback |
| Horizontal Scaling | Kubernetes HPA scales on consumer lag |
Stack: Java Spring Boot Apache Kafka Kubernetes Docker PostgreSQL Prometheus Grafana
A microservices-based KYC orchestration system that eliminates redundant verification across institutions while enforcing secure, consent-based data access.
The Model:
Institution A ──┐
Institution B ──┼──► [ KYC Service ] ──► [ Single Source of Truth ]
Institution C ──┘ │
│ Consent Gate
┌────▼─────┐
│ User │ ← Explicit authorization required
└──────────┘
| Design Principle | Implementation |
|---|---|
| Single Source of Truth | One verified identity, reusable across orgs |
| Consent as First-Class Primitive | No access without explicit user authorization |
| Event-Driven | Async service communication via Apache Kafka |
| Security-First | JWT + RBAC + encrypted storage |
| Lifecycle Automation | KYC renewal, revocation, and updates via workflows |
This system models trust boundaries and enforces controlled data propagation — it's a data governance system, not just a backend.
Stack: Java Spring Boot Spring Security Apache Kafka MySQL Redis Docker
A full-stack system that collects, normalizes, and analyzes competitive programming data across multiple platforms — with an insight layer, not just charts.
| Feature | Detail |
|---|---|
| Multi-Source Ingestion | REST + GraphQL pipelines across platforms |
| Insight Layer | Weakness detection, tag mastery, performance trends |
| Caching Strategy | 24-hour intelligent frontend cache to reduce API pressure |
| Scheduled Pipelines | Backend jobs continuously refresh analytics |
| Auth | OAuth 2.0 + JWT secure flow |
Stack: Java Spring Boot React.js Tailwind CSS Recharts MySQL Redis
Design for failure, not success
─────────────────────────────────────────────────────────
Decouple everything that can scale independently
─────────────────────────────────────────────────────────
Prefer async over blocking when correctness allows
─────────────────────────────────────────────────────────
Make state transitions explicit and traceable
─────────────────────────────────────────────────────────
Measure everything — latency, retries, failures
Building execution engines that remain correct under concurrency, failure, and scale.