PORTFOLIO / SELECTED WORK
Portfolio
My work on AI platforms and infrastructure.
These articles describe my part in each project and the decisions I made along the way.
Widearth: AR and digital twin platform
I handled deployment and operation of hybrid infrastructure and ML pipelines for a real-world AR service and digital twin platform. To support user-facing services alongside spatial-data processing, I separated their responsibilities and combined cloud and on-premise resources.
Read the articleOn-premise MLOps platform
I built an MLOps platform and GPU cluster to manage GPU-backed experiments and ML workloads as repeatable pipelines. On Kubernetes, I combined open source tools including Kubeflow and Argo Workflows to connect experimentation tools with the execution environment.
Read the articleHybrid Kubernetes cluster
I connected on-premise bare-metal Kubernetes with AWS EKS to use local resources while retaining cloud scalability and service continuity. I configured the cluster, workload execution, and status monitoring, using a Metric Collector and workflows to expand cloud workloads according to resource status.
Read the article