AI & Systems Engineering
Fourth-year undergraduate pursuing dual degrees in Data Science (SLIIT) and Physical Science (University of Ruhuna), engineering autonomous agent harnesses, hybrid RAG retrieval engines, and high-performance analytical data systems.
Production agent platforms, zero-dependency POSIX supervisors, and in-process columnar OLAP engines.
Led a 4-person engineering team to build an agentic dynamic pricing platform using Python, FastAPI, and LangGraph to collect market prices, optimize rates, and stream updates via Server-Sent Events (SSE); implemented SQLite tracking for per-user LLM token costs; contributed 376 commits as project lead.
Engineered a lightweight command-line task supervisor in Python to orchestrate detached background sub-agents; implemented fcntl.flock mutex locking and atomic file renames to prevent race conditions during concurrent multi-agent execution; built non-blocking lifecycle management (spawn, status, kill) with deterministic SIGTERM cascades and segregated logging streams.
Architected an episodic memory and hybrid retrieval engine combining local all-MiniLM-L6-v2 dense vector embeddings (FastEmbed) with BM25 keyword search over a multi-agent markdown knowledge base; built a unified REST API server for multi-agent state reconciliation, supporting atomic branch merges and cross-session knowledge retrieval.
Engineered a low-latency Linux speech transcription daemon binding directly to libwhisper.so (large-v3-turbo) via Python Ctypes; implemented a 150Hz 4th-order Linkwitz-Riley biquad filter and Silero ONNX streaming VAD with 400ms silence detection; built a non-focus-stealing PyQt6 Wayland desktop HUD.
Engineered an analytical intelligence platform over 263,000+ real retail transaction dockets across 8 years; built a Plan-Execute-Validate (PEV) agent orchestrator with PydanticAI to generate, execute, and verify SQL queries against a fast DuckDB analytical database; implemented driver-level read-only security boundaries and a React 19 assistant-ui interface with real-time SSE streaming.
Production frameworks, low-level runtimes, and systems infrastructure verified through implementations.
Quantified delivery milestones and formal higher education background.
Led engineering on FluxPricer (376 commits) with LangGraph state machine, cyclical error-recovery loops, and per-user token telemetry.
Engineered Plan-Execute-Validate (PEV) agent with PydanticAI over 263,000+ real retail transaction dockets in DuckDB with zero network latency.
Engineered SAM CLI in Python with fcntl.flock mutex locking and atomic swaps for detached background agent lifecycle management.
Built on-device speech transcription daemon binding directly to libwhisper.so via Python Ctypes with Silero ONNX streaming VAD.
BSc (Hons) in Information Technology — Data Science Specialization
2023 — Present (Expected 2027)
Undergraduate studies focusing on machine learning, multi-agent systems, information retrieval, and cloud computing architectures.
BSc — Physical Science
2024 — Present (Expected 2027)
Concurrent studies in pure mathematics, applied physics, and computational statistics.
Open for Full-Time AI Engineering, Systems Engineering, or Applied ML Internships.
📞 Phone: +94 71 929 6379
📍 Location: Malabe, Colombo, Sri Lanka