Joseph Miller · Infrastructure / Security / AI
I build, secure, recover, and modernize the systems people depend on.
I’m Joseph Miller, a hands-on systems engineer in Oregon. I’m drawn to migrations, failures, and inherited environments where the answer crosses virtualization, storage, identity, networking, cloud platforms, security controls, or recovery.
This portfolio is the record of work I have completed, systems I have built, and technical problems I have learned to make safer and more understandable.
Selected successful work
Virtualization · SAN · Recovery
Moved production workloads to Windows Server 2025 Hyper-V while preserving VM identity, networking, application availability, source SAN safety, and a recoverable starting point.
Evidence: Isolated boot, identity, storage, network, and application checks
Read the project →Private AI · GPUs · Agents
Built a high-memory professional NVIDIA GPU environment for local models, OpenAI-compatible APIs, agent gateways, model routing, and controlled remote access.
Evidence: Authenticated APIs, model routing, and controlled access
Read the project →Security monitoring · Forensics
Built repeatable Security Onion, Zeek, DShield, SiLK, PCAP, and IDS workflows for investigating activity across multiple evidence sources.
Evidence: Correlated packet, flow, log, and IDS evidence
Read the project →Containers · CI/CD · Security
Rebuilt this site around protected changes, multi-architecture images, non-root containers, health checks, runtime verification, and rollback.
Evidence: Protected PR, immutable image, runtime, and public-path checks
Read the project →Azure · PowerShell · Recovery
Deployed and repaired AVD session hosts, then turned the recovery sequence into staged scripts with discovery, tracking, registration, and validation logs.
Evidence: Discovery, registration, agent, profile, and session validation
Read the project →Data recovery · HFS+ · Linux
Recovered data from an old PowerPC-era disk image by identifying and repairing HFS+, mounting encrypted sparse images, and copying data to modern storage.
Evidence: Mounted the repaired source and copied recovered data safely
Read the project →What connects the work
A symptom in one system often begins in DNS, identity, storage, networking, certificates, tokens, virtualization, or time.
Before a risky change, I identify the authoritative source, the recovery point, the irreversible step, and the evidence that will prove the new state is healthy.
A running VM or green job is not enough. I check boot, identity, data, networking, applications, dependencies, and the outcome people actually need.
How I approach a hard system
Public engineering artifacts
This is a deliberately public, sanitized repository—not a feed from private systems or current infrastructure activity.
Security telemetry · public repository
A documented DShield sensor log collection and Elasticsearch ingestion workflow for security analysis.
Inspect the repository →What I’m building now
I’m designing an agent that can maintain a source-of-truth view of infrastructure, correlate telemetry and prior incidents, explain likely failure paths, recommend fixes with evidence, and verify recovery.
The boundary is architectural: source access is read-only, remediation stays human-controlled, and nothing from the assistant is exposed through this public site.
My environment for infrastructure, monitoring, local AI, recovery, and agent experiments that I can test directly and document honestly.
See the lab