See the whole system
A useful product starts with the relationships between people, data, infrastructure, and the interface - not with a framework choice.
ngayantha.comTechnology · Ideas · SystemsStart exploring
Software engineer at Paraqum Technologies, building scalable full-stack applications, real-time data products, and interfaces with a strong eye for design.
01 - The way I work
Software engineer
Technology enthusiast
End-to-end builder
My work moves from architecture and APIs to cloud environments and the interface people eventually touch.
An honours degree in Information Technology gave me a broad view of software and information systems. Production engineering sharpened it: every decision has to survive real users, changing requirements, live data, and the next person who reads the code.
Professional work has moved me from implementation into architecture, ownership, and technical decision-making across complex software products. Independent builds and research keep that experience connected to full-stack systems, cloud technology, and applied AI.
A useful product starts with the relationships between people, data, infrastructure, and the interface - not with a framework choice.
Good engineering reduces the effort required to understand, operate, and extend a system without hiding what matters.
Leadership means making the problem, trade-offs, and direction clear enough for other people to contribute with confidence.
Good teamwork keeps decisions visible, feedback useful, and ownership shared from the first discussion through delivery.
02 - The route so far
A compact record of the moments that changed the scale of the problems I could take on.
Foundation
Began an honours degree at the University of Kelaniya, building a base across software, data, information systems, and research.
Production
Began working professionally as a software engineering intern and learned how technical decisions behave beyond the classroom.
Ownership
Progressed into broader responsibility across complex interfaces, real-time data, APIs, and product architecture.
Now
Graduated with a 3.65 GPA, published two IEEE papers, and moved into a Software Engineer role spanning platforms, cloud, data, and product delivery.
03 — Experience
Paraqum Technologies · Sri Lanka
04 - Selected research
A focused study of how machine-learning methods can support DDoS detection and mitigation in programmable network environments.
Publications
05 — Education
University of Kelaniya · Sri Lanka
2021—25
GPA / 4.00 · Second Class Upper
Dean's List for academic performance
06 — Selected work
A selection spanning full-stack products, institutional systems, commerce, and applied machine learning.
React · NestJS · PostgreSQL · Prisma · Docker
Architecting a full-stack platform for charitable donations and community fundraising, with a modular backend for secure transactions and user roles.
2026 — NowDeep learning · CNN · Data augmentation
A deep-learning object-detection pipeline for Melbourne University, Australia, using data augmentation and hyperparameter tuning to optimise CNNs for high-speed, real-time inference.
2024Next.js · React · Strapi · PayHere
A scalable storefront connecting a modern interface, headless content, and local payment infrastructure.
2025Next.js · React · Microsoft 365 Lists
A focused portal for securely storing and validating employee training certificates.
Sep 2025React · Node.js · Express · MySQL
A management platform created for Sri Lanka Law College using an iterative Agile delivery process.
Mar 202407 - Working stack
Interfaces, services, infrastructure, and data—chosen for the problem, not the trend.
A practical toolkit for moving from a precise interaction to the systems behind it, without losing sight of speed, maintainability, or the person using the result.
InterfaceReact
LanguageTypeScript
FrameworkNext.js
LanguageJavaScript
InterfaceTailwind
ArchitectureREST APIs
WorkflowGit
SystemsLinux
DesignFigma
MotionGSAP
RuntimeNode.js
LanguageJava
LanguagePython
DataPostgreSQL
DataMongoDB
InfrastructureDocker
CloudAWS01
02
03
AWS Certified Cloud Practitioner in progress.
08 - Writing
A three-part series, “Building AI systems that hold up in production”: MCP, agentic retrieval, and schema-first APIs, explained for the decisions that matter after the demo.
All writing
An open protocol can make your integrations reusable. The interesting work starts when you decide what those integrations are allowed to do.

Hybrid search, rank fusion, and reranking can improve your evidence. Production retrieval also needs to respect the parts of a question that aren’t negotiable.

How to combine TypeScript, runtime schemas, authorization, and database guarantees without pretending one of them can do every job.