Active project
Artificial Intelligence
Local Qwen inference server
Running an open-weight coding model on an iMac Pro, with llama.cpp inference and an Open WebUI interface.
PROJECT / 01
ENGINEERING PORTFOLIO & TECHNICAL LAB
Engineering
systems.
Building software.
Exploring AI.
I’m a Media Systems Engineer working across broadcast technology, networks, and infrastructure. Drabsnarg is where I document what I build, what I learn, and what comes next.
REAL SYSTEMS. DOCUMENTED DECISIONS. ONGOING EXPERIMENTS.
DRABSNARG / SYSTEM ARCHITECTURECONCEPT · 01
Public portfoliodrabsnarg.com
PUBLICIndependent services
PRIVATE AI ENVIRONMENT
Secure accessCloudflare Access + Tunnel
Open WebUIIndividual accounts · chat interface
Local inferencellama.cpp → Qwen
iMac Pro / 128 GB RAM / Owner-operated
Systems engineeringBroadcast & mediaNetworkingSoftwareLocal AI
BUILT TO KEEP EVOLVING ↗SYSTEM MAPS / THE SIGNAL PATH
There is always another layer.
- 01Authenticated user
- 02Cloudflare Access + Tunnel
- 03Private chat interface
- 04Local inference service
- 05Open-weight model
- 01Source / camera
- 02Transport: SDI or NDI
- 03Production / processing
- 04Encoder / delivery
- 05Monitoring + diagnostics
- 01Source + changes
- 02Automated checks
- 03Versioned release
- 04Staged deployment
- 05Health verification
01 /SELECTED WORK
From idea to implementation.
Planned
Home Lab and Infrastructure
Secure access to Drabsnarg AI
An invitation-only access design connecting Cloudflare Access and Tunnel to the existing local AI stack.
PROJECT / 02
Planned
Software Development
Auditable AI coding workflows
A planned study of coding assistants, independent code reviews, and controlled multi-agent development workflows.
PROJECT / 03
Project labels distinguish active work from planned experiments. Results appear only when documented.
02 /TECHNICAL DISCIPLINES
Connected by curiosity.
My work sits between disciplines. Understanding the connections is often where the most useful engineering happens.
01
Broadcast & media systems
Production infrastructure, signal transport, audio networking, and the systems behind live media.
02
Networks & infrastructure
Routing, switching, segmentation, and the connections that make complex systems work together.
03
Software & automation
Practical tools, Python and PowerShell workflows, APIs, and repeatable solutions to real problems.
04
AI & local inference
Open-weight models, owner-operated hardware, inference experiments, and AI development tools.
05
Systems administration
Windows, macOS, servers, Microsoft 365, identity, device management, and reliable operations.
06
Security & experimentation
Secure access, infrastructure ownership, and a systematic approach to learning how things work.
THE DRABSNARG AI LAB
Local hardware.
Open models.
My own infrastructure.
An evolving environment for self-hosted AI, model experiments, and software development. Built around an iMac Pro, llama.cpp, and Open WebUI.
AI APPLICATION: INVITATION-ONLY · AUTHENTICATION REQUIRED
Owner-operated inference128 GB iMac Pro · Qwen coding model
A deliberate access boundaryLocal stack tested. Secure remote access planned.
Experiments with contextDocument the configuration, method, and limitations.
04 /THE ENGINEERING NOTEBOOK
Notes from the workbench.
Oct 26, 2026Experiment
AI & Local Inference
Local LLM Benchmark Method
A repeatable plan for comparing local coding models without invented numbers.
Oct 26, 2026Technical reference
Broadcast & Media Systems
Investigating a PTP Clocking Problem
A structured way to distinguish clock election, network reachability, and audio-flow symptoms.
Oct 26, 2026Technical reference
Software & Automation
Safer PowerShell Deployment Checks
Preflight validation, exit-code checks, and post-deployment smoke tests.
KEEP EXPLORING
There’s always another layer.
See the current focus, explore the technical notes, or learn more about the engineering behind Drabsnarg.