Hello, I'm Dineth
I'm Dineth Hettiarachchi, an AI/ML systems engineer building end-to-end systems across machine learning, data engineering, software, and edge computing.
My work connects models, data, and hardware. I build MLOps and analytics pipelines, GPU and TPU architecture simulators, embedded IoT platforms, and applied AI systems that are understandable, testable, and useful beyond the demo.
Portfolio · LinkedIn · X · Medium · Substack
- Parallel and accelerator computing: TinyGPU makes SIMT execution, memory, and synchronization visible in Python; TinyTPU runs a SystemVerilog systolic array in the browser through WebAssembly.
- Machine learning and data engineering: the Telco Churn MLOps Pipeline combines MLflow, Spark, Kafka, Airflow, and Dockerized inference; the Airbnb Market Intelligence Pipeline focuses on reproducible analytics with DuckDB, Parquet, data-quality checks, statistics, and machine learning.
- Edge and applied AI: the Smart Beehive Monitoring System connects solar-powered ESP32 sensing to Firebase and a React/TypeScript dashboard; the Multi-Omics Cancer Subtype Classifier applies XGBoost, PyTorch, biological pathways, SHAP, and integrated gradients to cancer subtype classification.
AI/ML and MLOps · Data and software systems · Parallel and edge computing
Python, PyTorch, Apache Spark, FastAPI, Docker, TypeScript, C++, and SystemVerilog.
I write about practical AI/ML systems, parallel computing, and embedded engineering on Medium and Substack. For collaboration, engineering opportunities, or technical discussion, contact me through LinkedIn or email.