name: Rohan Silva
role: Software Engineer II – DevOps
company: H2O.ai
team: Predictive AI Engineering
location: Sri Lanka
experience: 3+ years
focus:
- Cloud Infrastructure
- CI/CD & Release Automation
- Security Scanning & Compliance
products:
- H2O-3
- H2O-3 Secure
- Driverless AI
what_i_do:
- Design and automate infrastructure across multi-cloud environments
- Build and maintain CI/CD pipelines, nightly tests, and release workflows
- Publish marketplace images and coordinate secure releases
- Optimize cloud costs with Spot, right-sizing, and resource management
- Automate CVE scanning and enforce infrastructure access controls
- Troubleshoot complex failures across distributed systems
philosophy: >
Solve infrastructure problems at their root,
turn repetitive manual tasks into reliable automation,
and build systems that help engineering teams ship
with confidence and efficiency.|
Analyzed AWS infrastructure spending across EC2, Jenkins and storage; identified cost-saving opportunities and evaluated Spot Instances, storage reductions and Savings Plans. Impact: Clear cost visibility and a data-driven roadmap for reducing cloud expenditure. |
Developed and standardized publishing workflows for Driverless AI VM images across AWS, GCP, Azure and OCI, using reusable GitHub Actions, HCL2 and Terraform-based validation. Delivered the release images to AWS, GCP and OCI. Impact: Consistent multi-cloud releases and automated image validation. |
|
Implemented Spot-first execution with cross-AZ fallback and On-Demand recovery. Analyzed 30 days of CI runs to identify capacity shortages, interruptions and opportunities to improve Spot utilization. Impact: Better capacity recovery and measurable visibility into CI compute cost and reliability. |
Replacing oversized, manually maintained CI images with automated CPU/GPU AMI builds for x86_64 and ARM64, including current Ubuntu and CUDA/NVIDIA tooling. Impact: Reproducible runner images, simpler maintenance and less wasted storage. |
|
Delivered the official H2O-3 Secure releases end to end: release pipelines, automated jobs, testing and publication coordination. Impact: Dependable delivery of official Secure releases. |
Improved GitHub Actions and Jenkins engineering workflows, reusable automation, nightly testing, smoke tests and release processes across Driverless AI and H2O-3. Impact: Less repetitive manual work and more consistency across engineering workflows. |
|
Built and maintain Driverless AI vulnerability-scanning workflows with nightly scan version tracking, plus GitHub OIDC authentication and AWS IAM permission design. Impact: Maintainable security workflows and controlled, auditable infrastructure access. |
Investigated an EKS architecture with GitHub Actions Runner Controller (ARC) and Karpenter, alongside additional Availability Zones, Spot placement scoring and alternative capacity strategies. Impact: Concrete paths toward scalable, resilient and cost-efficient CI infrastructure. |
