Anthony A. Robinson

AI Systems Architect & Senior DevOps Engineer · Redmond, WA

Senior DevOps with 20 years of Microsoft-scale build/release rigor, now running production AI pipelines where LLMs do what humans used to — Claude, OpenAI, or Copilot, whichever the job needs.

The wedge

Production AI ops, not AI hobbyist.

Daily-driven production AI pipelines doing real work at SanMar — meeting intelligence, executive activity reports, ADO work-item reconciliation. Built on Anthropic and OpenAI APIs, with the right model picked for each step. Retry/backoff, cost models, and observability — not just a chat window.

Twenty years of build/release rigor, applied to LLMs.

The same discipline that ran 100+ Windows Mobile builds and scaled a Service Pack process from 4 to 26 languages now runs prompt evaluation harnesses, anti-drift safeguards, and per-project memory persistence across 22 active Claude Code projects.

Specification precision.

I write mega-prompts the way I used to write build scripts: explicit escalation tiers, output format contracts, documented anti-patterns. Five production-grade mega-prompts running today, each 230+ lines, each holding the line against AI drift.

Numbers that survive compression

233
PowerShell modules in the composable AI context library
22
active Claude Code projects with documented execution plans
13
category meeting classification taxonomy
150+
scoped tool permissions governing AI agent access
97%
Azure deployment time savings (redesigned pipeline)
48%
serial-to-parallel pipeline time savings

Proof

TLDW
AI content intelligence: 51 YouTube channels, 783+ videos, automated transcription → classification → summarization. Active daily pipeline.
Transcripts Pipeline
Audio → Whisper → Haiku classification → Sonnet summarization → ADO work-item reconciliation. Dual-model routing by task complexity.
Homelab Infrastructure
30+ Docker Compose services, Kubernetes on Proxmox/Talos with OpenTofu + Flux GitOps, SOPS-encrypted, MetalLB + Traefik.
AI Systems Competencies
The seven competencies — specification precision, evaluation, multi-agent orchestration, failure patterns, trust/security, context, token economics — with one concrete example each.

The full record

Same content, three formats. Pick whichever your filter prefers.

Resume (HTML) Resume (JSON) Resume (Word) For LLMs (llms.txt)

Contact

[email protected] +1 206 979 9942