A production-grade, single-binary CLI tool for zero-downtime database migrations written in Go. Features streaming pipelines (no disk I/O), automatic FK dependency resolution via topological sort, a Dead Letter Queue for resilient failure handling, and a built-in PII auditor.

192,493 rows/sec — MySQL to PostgreSQL, 200k row dataset

Tech Stack: Go PostgreSQL MySQL MongoDB SQLite Streaming Pipelines

View Repository →

A complete 7-service microservices backend for e-commerce. Implements an event-driven SAGA choreography pattern to guarantee distributed transaction consistency without 2PC overhead. Features a Kong API Gateway, Eureka service discovery, and per-service PostgreSQL databases.

Zero 2PC overhead — SAGA choreography over Kafka for resilient distributed transactions

Tech Stack: Spring Boot Java Kafka Docker PostgreSQL Redis Kong API Gateway

View Repository →

A full-stack event management platform with QR-code-based check-in, dynamic seat allocation, and real-time capacity tracking. Built to handle concurrent bookings without race conditions using optimistic locking and transactional guarantees.

Concurrent-safe bookings — optimistic locking with zero double-allocation

Tech Stack: Spring Boot Java PostgreSQL React QR Scanning WebSocket

View Repository →

A book search and discovery app for Android, iOS, and Desktop from a single Kotlin codebase using Kotlin Multiplatform. Clean architecture with shared business logic and Compose Multiplatform UI.

One codebase → 3 platforms — shared UI and business logic via KMM

Tech Stack: Kotlin Jetpack Compose KMM Ktor SQLDelight Clean Architecture

View Repository →

An offline-first Android chat application with real-time message delivery. Uses GraphQL Subscriptions via Apollo for push-based instant messaging and Room DB for local persistence — messages remain available without internet.

Offline-first — Room DB sync ensures zero data loss on reconnect

Tech Stack: Kotlin Jetpack Compose GraphQL Apollo RoomDB MVVM

View Repository →

An AI-powered financial agent using a Retrieval-Augmented Generation pipeline. Queries a corpus of financial news articles via a vector database to answer domain-specific questions with grounded, source-cited responses.

Grounded responses — RAG over vector DB eliminates hallucinations on financial data

Tech Stack: Python LangChain LLMs Vector DB FastAPI RAG

View Repository →