I build backend systems that hold up under real traffic — and I write about how I built them.
Over the past year as a backend developer, I've shipped more than twenty production features, cut a page's load time from 25 seconds down to 5, and migrated a monolithic scheduler into an independent microservice — all while learning the Spring ecosystem on the job. Outside of work, I'm finishing a Master's in Computer Science, researching anomaly detection for cybersecurity, and rebuilding an Order Management System the way I'd want to find one in production: cleanly decomposed, event-driven, and deployed end to end.
Experience
My biggest win here was performance: I tracked down why a core page was taking 25 seconds to load and brought it down to 5 seconds by optimizing MySQL queries and refactoring the Quarkus service layer — a fix that directly improved the daily experience of production users.
Over the year I shipped 20+ full-stack features across two live client-facing applications — POS, sales, voucher management, barcode generation, and real-time WebSocket communication — using Quarkus, Vue.js, and MySQL, presenting demos directly to clients in Agile sprint cycles.
I also led the migration of a monolithic scheduling system (Alert Wizard) into an independent microservice using Quarkus and Docker, and coordinated zero-downtime production releases — building Docker images, writing Bash deployment scripts, and working closely with QA and system admins for stable rollouts.
I built a complete Futsal Management System independently — slot scheduling, court booking, and Khalti payment integration — iterating through direct supervisor feedback from design to deployment. I also contributed production-ready backend features and REST APIs to a microfinance staff training module, joining client demos, sprint meetings, and Docker-based deployments on WildFly.
Research
I design Python-based data pipelines for cybersecurity log ingestion and APT (Advanced Persistent Threat) detection, implementing anomaly detection models using Hidden Markov Models (HMM) across large-scale network datasets.
Projects
A production-style OMS built with Spring Boot and Spring Security, covering the full order lifecycle with JWT-based authentication and role-based access. I decomposed the monolith into 6 microservices — API Gateway, Order, Account, Payment, Identity, and Service Discovery — using Spring Cloud with Kafka for event-driven communication, then containerized everything and deployed it to AWS with Kubernetes.
A RAG chatbot built with Spring Boot, Spring AI, OpenAI's API, and Ollama (Llama 3.2), letting students query university information in plain language instead of digging through the website. I built the full ingestion pipeline myself — scraping with Jsoup, chunking text, and storing embeddings in PostgreSQL with pgvector for accurate, context-aware answers.
Tools & technologies
Writing
A walkthrough of how I designed AskTiger end to end — from scraping and embeddings to wiring Spring AI and pgvector into a working RAG chatbot.