Production AI infrastructure · Cloud-native data systems · Robotics & autonomous perception

Website · LinkedIn · Email · Red Hat Developer · Writing · Speaker profile

I am a Software Engineer at Red Hat AI, building production data-processing infrastructure on Kubernetes and OpenShift. I work across Apache Spark, Kubeflow, distributed document processing, secure multi-tenant operators, and observability.

An M.S. in Robotics & AI and hands-on work in LiDAR, computer vision, autonomous systems, and GPS/IMU sensor fusion give me a robotics foundation for designing observable, failure-aware systems.

I am a reviewer for the Kubeflow Spark Operator, contributing code, design, and reviews across deployment, reliability, security, multi-tenancy, and scheduling.

Starting with making Kustomize a complete, first-class deployment method alongside Helm, I built and validated the deployment path with the same Go E2E suite and added semantic drift checks to keep both packaging methods aligned. From there, I tightened least-privilege RBAC, hardened controller-runtime caches against memory-exhaustion failures, introduced namespace-scoped webhook controls for multi-tenant clusters, and improved Kubernetes Event reliability.

At Red Hat AI, I extended that work into OpenShift AI and Open Data Hub by moving Spark Operator integration to an out-of-tree module, adding resilience and E2E coverage, and building configurable namespace scope, Prometheus access, and operator/webhook NetworkPolicies. My current upstream work focuses on Kubernetes-native workload-aware scheduling, reliable events and observability, and a stronger Spark/Kubeflow developer experience.

Explore my ongoing open-source work: PRs I have opened · PRs I have reviewed

I help shape SparkClient's production API through design review and community work. I am also a CNCF LFX mentor for Evolving SparkClient into Kubeflow's Unified Data Processing Layer—extending the path from raw data to trained models with observability, scheduled and streaming execution, and Spark-to-Trainer workflows.

Programming: Go, Python, C++, JavaScript, SQL, MATLAB
Cloud-native + MLOps: Kubernetes, OpenShift, Kubeflow, Docker, Helm, Kustomize, MLflow, GitHub Actions, AWS SageMaker, Amazon Bedrock, Azure OpenAI
Data, ML + perception: Apache Spark, PyTorch, TensorFlow, scikit-learn, Pandas, NumPy, OpenCV, Point Cloud Library (PCL), Docling, LiDAR, sensor fusion
Systems + tooling: Kubernetes Operators, controller-runtime, FastAPI, REST APIs, PostgreSQL, Git, Linux, Prometheus, Selenium, E2E/integration testing

Current interests: Kubernetes-native workload scheduling, secure multi-tenant AI platforms, observable distributed data processing, Spark/Kubeflow developer experience, and perception systems that combine geometry with learned models.

• Co-authored 5 anti-patterns that cause Kubernetes operator vulnerabilities and Protect your Kubernetes Operator from OOMKill for Red Hat Developer, turning upstream cache hardening and broader controller audits into practical production guidance.
• Write about distributed Docling + Spark, ML, computer vision, and system design on Medium.
• Mentor contributors through CNCF LFX and the Kubeflow community; find my speaker profile and technical videos online.

If you are working on cloud-native AI, distributed data processing, or autonomous systems, reach me through LinkedIn or email.