- Re-architected core software to seamlessly handle 200 million transactions per second, excelling under high-pressure conditions in C# with Kafka and GaussDB (Postgres SQL).
I architect scalable web & desktop systems in C# and the cloud — from pipelines moving 200M transactions/sec to open-source distributed AI.
A full-stack engineer with a decade architecting scalable web & desktop applications — specialising in C#, cloud infrastructure, applied AI and blockchain.
Across ten years I've moved from migrating legacy VB6 estates to .NET, to engineering ASP.NET APIs under extreme load, to deploying machine-learning models for facial and text recognition in production.
Most recently at Huawei I re-architected software to handle 200 million transactions per second in C# with Kafka and GaussDB — and I've shipped blockchain validation systems, distributed AI on actor models, and led engineering teams along the way.
I care about systems that stay fast and correct under pressure, clean architecture, and contributing back to open source.
An open-source CLIP (Contrastive Language–Image Pre-training) implementation in C#, using a Microsoft Orleans actor-model approach to build a fully distributed AI mechanism.
Engineered FxLinq to filter, sort, group, project and aggregate any IEnumerable<T> using Microsoft Power Fx — the formula language behind Power Apps, Power Automate and Dataverse — through a single extension-method call.
Leverages SAM 2 for precise background extraction and seamless white-background rendering. Uses OpenAI CLIP and YOLO for prompt-driven image classification — each image is categorised against a curated prompt set and routed to output folders labelled by predicted class.