bits nor bytes

by Sachira Chinthana Jayasanka

About me

A software engineer born in Sri Lanka, living in the United States at the moment. I have spent the last 12+ years building software across backend systems, cloud platforms, mobile applications, and internal developer tooling. Over time I have worked with .NET, Java, React, React Native, AWS, and Azure, while keeping a strong interest in system design and the messy practical side of making software work well in production.

Most recently I worked at Amazon as a Software Development Engineer, where I spent nearly five years building internal platforms and tools across a wide range of problems. That work included large migration efforts, BI data extraction tooling for leadership, new sensor integrations for the Alexa app, developer tooling for macOS/iOS/iPadOS engineers, and a multi-phase migration from QLDB to DynamoDB. A big part of that role was writing design documents, coordinating with multiple teams, and turning complicated operational problems into systems that were easier to scale and maintain.

Before that I worked at ImageWork Technologies as a Senior Software Engineer. That role pushed me deeper into full-stack and cloud-heavy work: building Web APIs for SPAs, implementing authentication flows, designing long-running .NET services, and using Azure Web Jobs, Functions, and Azure DevOps pipelines to support secure multi-environment deployments. I also worked on large-scale document processing systems, including automation flows that handled more than a million documents.

At this point I see myself as a full-stack engineer who is most comfortable when there is a system to design, a rough edge to smooth out, or a workflow to make more reliable. I still enjoy working close to both backend and frontend layers, but I am especially drawn to architecture, platform design, and building tools that make other engineers more effective.

Projects

Workflow Nexus – AI-Native Orchestration Platform

Website: https://arandomsitein.space/

Architected and built Workflow Nexus, an event-sourced execution engine for orchestrating multi-step AI content generation workflows combining LLM reasoning, user interaction, and media synthesis through declarative configuration.

Developed using an AI-native engineering workflow: the entire codebase and repository architecture documents were AI-assisted under my direction. All designs were reviewed, validated, and refined by me prior to implementation. I owned system architecture, integration strategy, and production hardening.

System Design & Execution

  • Designed event-sourced state model (MongoDB-backed immutable log, replayability, checkpoint recovery, auditability)
  • Built modular execution framework with pluggable step types (API, transformation, interaction, media)
  • Implemented actor-based worker system with queue persistence, concurrency controls, heartbeat monitoring, and fault recovery
  • Enabled real-time execution streaming via SSE

Interface & Extensibility

  • Engineered schema-driven UI architecture (JSON Schema → dynamic React rendering)
  • Created shared component system and headless terminal interface supporting the same protocol
  • Abstracted multi-provider AI integrations (OpenAI, Anthropic, MidJourney API, Leonardo.ai, ElevenLabs, local Stable Diffusion) behind a unified contract layer
  • Dual storage strategy (structured data in MongoDB, media on filesystem with references)
  • Versioned migrations and containerized deployment (Docker, Nginx)

Infrastructure

  • Dual storage strategy (structured data in MongoDB, media on filesystem with references)
  • Versioned migrations and containerized deployment (Docker, Nginx)

Repository: https://github.com/SachiraChin/workflowmanager-showcase

dotnet-monitor-ui

A tool designed to work with APIs provided by dotnet-monitor tool. This application does not have a server-side, purely works on the browser to access APIs from dotnet-monitor. The reason I thought this tool would come handy was when I was watching session by Sourabh Shirhatti where he was explaining functionalities provided by dotnet-monitor. The current version of the app has the functionalities to access all endpoints of dotnet-monitor and on-demand management of /metrics endpoint. Please read about dotnet-monitor here.

Repository: https://github.com/SachiraChin/dotnet-monitor-ui

SharePoint authentication for modern web apps

SharePoint.Authentication is an inject-able SharePoint context and token helper which can be used in multi-tenant applications. Abilities of this library extended by its sister library SharePoint.Authentication.Owin.

Reason I came up with this project is due to problems I’ve met while creating high trust multi-tenant application. In the project I’ve worked on had different client id and secret for each tenant. As you may recall, SharePoint context provider automatically added to web project currently rely on only one client id and secret which must be added to web.config file. This was not the solution I wanted because, application had different client id and secret (provided by seller dashboard) for low trust part of the app and had different client id and secret for each tenant for high trust. To authentication layer to function properly, I wanted it to instantiated per tenant/user and wanted it to inject-able via Unity container. This is the solution I came up with to fix that issue.

Solution I came up consists of two separate libraries, SharePoint.Authentication and SharePoint.Authentication.Owin. First library is the base which I made from token helpers and providers made by Microsoft. Second library made with one specific purpose in mind which is to use SharePoint authentication in MVC or Web API based applications. SharePoint.Authentication contains all the functions you need to build an authentication layer, but if you want don’t want to build one yourself, you can use SharePoint.Authentication.Owin.

Repository: https://github.com/SachiraChin/SharePoint.Authentication