Place workloads deliberately
Choose between cloud, local, and hybrid environments based on data, latency, resource, and operational requirements.
The foundation for useful systems
Cloud & Infrastructure
Infrastructure built to run—and be understood.
Infrastructure is a set of decisions about where work runs, how systems communicate, and what happens when a component fails. I approach it from the workload outward, balancing capability, cost, security, and the effort required to operate the environment over time.
Explore the capabilities- Cloud architecture
- Networking
- Virtualization
- Local AI compute
- Platform operations
01
Problems worth solving
Connect environments reliably
Make routing, segmentation, remote access, and name resolution understandable and maintainable.
Keep systems recoverable
Define backups, recovery objectives, monitoring, and ownership before a failure forces those decisions.
02
Explore the capabilities
01Cloud & hybrid architecture
Map application and data needs onto an environment with clear network, identity, availability, and cost boundaries.
Methods & tools
- AWS, Azure, and Google Cloud
- Hybrid connectivity
- Workload placement
- Cost analysis
Example deliverables
- Architecture diagram
- Platform tradeoffs
- Deployment plan
02Networking & connectivity
Design addressing, routing, segmentation, and remote access as a coherent whole, with a clear path for troubleshooting.
Methods & tools
- VLANs and routing
- DNS
- VPN connectivity
- Firewalls
Example deliverables
- Network plan
- Traffic and trust map
- Troubleshooting runbook
03Virtualization, containers & storage
Build repeatable environments and match storage behavior to application needs. Account for failure domains and lifecycle management.
Methods & tools
- Virtual machines
- Containers
- Storage design
- Backup and recovery
Example deliverables
- Environment topology
- Capacity plan
- Recovery procedure
04Local AI & research compute
Plan compute around model size, memory, data movement, and concurrent workloads. Validate actual throughput rather than relying on hardware specifications alone.
Methods & tools
- GPU workloads
- Model serving
- Resource profiling
- Experiment environments
Example deliverables
- Workload requirements
- Resource measurements
- Serving architecture
05Platform operations
Make deployments, monitoring, patching, and incident handling repeatable so the environment remains understandable after the initial build.
Methods & tools
- Deployment pipelines
- Observability
- Configuration management
- Lifecycle planning
Example deliverables
- Operational baseline
- Deployment workflow
- Ownership and maintenance plan
03
From question to working system
- 01
Map the workloads
Identify applications, data, users, and the service levels they need.
- 02
Compare the tradeoffs
Evaluate placement, cost, security, reliability, and management effort.
- 03
Build the foundation
Establish network, identity, storage, and repeatable deployment paths.
- 04
Prove recovery
Validate monitoring, backup restoration, and the handover to ongoing operations.
Start with what the environment needs to support.
Discuss the workloads, constraints, and operational problems behind your next infrastructure decision.