Software Engineer · AI & Agentic Systems · Backend & Distributed Systems
I'm Oshan Ranasinghe, a Software Engineer focused on backend, distributed, and AI-enabled systems. I design and build agentic applications, multi-agent architectures, asynchronous workflows, APIs, and observable distributed systems.
My work spans professional backend and R&D engineering alongside independent projects in agentic AI, distributed systems, and intelligent-agent research.
Selected engineering work
These projects represent the intersection of my software engineering, distributed systems, and AI work — from research-grade multi-agent communication to applied agentic pipelines.
Flagship Research · Distributed AI
An ontology-based multi-agent system for enabling heterogeneous intelligent agents to detect semantic conflicts, communicate, and negotiate shared meaning.
Built around distributed agent communication, service discovery, negotiation workflows, persistence, and LLM-assisted semantic reasoning.
Open Multi-Agent Framework
An open, domain-agnostic multi-agent framework for decentralized collaboration between specialized autonomous agents.
v0.1 is design-complete and implementation is pending. The architecture defines parent-driven routing, broker-mediated communication, private agent state, shared task state, memory and knowledge boundaries, and first-class observability.
Applied AI · Personalization Platform
A personalized knowledge-discovery platform combining content discovery, explainable ranking, feedback-driven user modeling, and agentic adaptation.
Implemented with a FastAPI backend, layered application services, PostgreSQL/pgvector, RabbitMQ/Celery, traceable recommendation scoring, adaptive agents, observability, tests, and CI.
Professional engineering
CodeGen International · R&D Engineering
Worked on Java and Python distributed agent systems, including end-to-end OpenTelemetry tracing across a 10–12+ agent orchestration framework visualized through Jaeger and Grafana, alongside RabbitMQ messaging, gRPC communication, RAG/Weaviate work, and containerized engineering workflows.
Treeloq · formerly Priwoo
Built customer-facing backend and full-stack application features using Node.js, Express.js, MongoDB, React, REST APIs, JWT authentication, third-party integrations, AWS Lambda, and PDF-generation workflows with Puppeteer.
Engineering capabilities
Additional projects
A two-stage research system that reconstructs PDFs into structured Markdown/JSON, then runs a LangGraph analysis workflow with explicit claim verification and hallucination controls.
A multi-user conversational AI backend with three-layer memory, conflict-aware long-term facts, SSE streaming, LLM self-evaluation, JWT authentication, and distributed tracing.
An architecture-oriented collaborative editing system exploring real-time communication, persistent document state, operation propagation, and service boundaries.
Engineering direction
My longer-term engineering direction extends toward autonomous and embedded systems, where AI software meets real-time computing and physical systems. I'm currently strengthening the C++ and systems foundations needed to move deeper into that space while continuing to build production-oriented software and AI systems.
Open to opportunities
I'm currently open to Software Engineering, AI Engineering, Applied AI, Agentic AI, and AI/backend engineering opportunities in Sri Lanka and internationally.