Anthony A. Robinson
Professional Summary
AI Systems Architect and Senior DevOps Engineer with 20+ years building enterprise-grade build, release, and deployment infrastructure at Microsoft and beyond. Now designing, building, and operating AI-augmented automation systems that integrate large language models into production DevOps workflows. Proven expertise across all seven core AI systems competencies: specification precision, evaluation design, multi-agent orchestration, failure pattern recognition, trust/security architecture, context engineering, and token economics. Builder of production pipelines where Claude, Whisper, and custom AI tooling replace manual knowledge work at enterprise scale.
AI Systems Architecture Competencies
Specification Precision
Authored 5 production-grade AI mega-prompts (230+ lines) with anti-drift safeguards, explicit escalation tiers, and output format contracts. Designed a 13-category meeting classification taxonomy with boundary detection rules for multi-meeting transcripts. Created Copilot agent instruction sets with enforced naming conventions, directory organization patterns, and documented anti-patterns to prevent AI drift.
Evaluation & Quality Judgment
Built a structured PowerShell security audit framework covering 6 risk categories (credential exposure, exfiltration, dynamic code execution, Base64 obfuscation, temp file races, network MITM). Developed a 27-point quality checklist for AI-generated visual artifacts with mandatory render-validate-fix loops. Enforced README accuracy policies requiring review after every substantive code change.
Multi-Agent Decomposition & Delegation
Operate 22 active Claude Code projects with 42 documented execution plans. Architect multi-model pipelines using Claude Haiku for high-throughput classification and Claude Sonnet for deep analysis/generation — selecting models by task complexity and cost profile. Design planner-agent workflows with parallel sub-agent dispatching for independent workstreams.
Failure Pattern Recognition
Created an ADO API resilience framework addressing transient network failures with progressive retry-backoff (5 attempts, 2–10s delays), per-item isolation to prevent cascade failures, and ADO pipeline-native error logging. At Microsoft, designed automated test-failed-retry pipelines to isolate flaky infrastructure from genuine failures. Mitigate context degradation through session boundary management and specification reinforcement.
Trust & Security Design
Architect blast-radius-aware permission systems with 150+ scoped tool permissions governing AI agent access to infrastructure. Implement human-in-loop approval gates for deployment operations. Integrate Azure Key Vault for secret management across 8+ automation modules. Built security scanning for PowerShell scripts covering exfiltration vectors, backdoor patterns, and credential leaks.
Context Architecture
Designed and maintain a 233-module composable PowerShell library serving as a reusable context layer for AI-assisted development, with a "search-before-create" protocol enforced via global CLAUDE.md directives. Implement per-project memory persistence files for cross-session state tracking. Architect dual-file documentation patterns (raw transcript + structured summary) linking meeting insights to Azure DevOps work items for continuous context flow.
Cost & Token Economics
Built dynamic token allocation modules that adjust API limits based on content length and task complexity. Operate blended-cost model strategies routing high-volume classification to Haiku (~$0.25/M tokens) and complex analysis to Sonnet, optimizing spend across production pipelines. Calculate and forecast token burn for batch operations across hundreds of transcripts and reports.
Key AI/ML Skills
| LLM APIs & Platforms | Anthropic Claude API (Haiku, Sonnet, Opus), OpenAI API, Claude Code, GitHub Copilot, Cody, Cline |
|---|---|
| AI Engineering | Prompt engineering (mega-prompts, specification precision, anti-drift), multi-agent orchestration, evaluation harnesses, token economics |
| ML & Audio | Whisper (GPU-accelerated transcription, CUDA, batch processing), faster-whisper |
| Automation & Orchestration | n8n (workflow automation), FastAPI, PowerShell module composition (233 modules) |
| Infrastructure | Docker, Kubernetes (Flux GitOps), Terraform/OpenTofu, Proxmox, Talos Linux, Ansible |
| Cloud & DevOps | Azure DevOps, Azure CLI, Azure Key Vault, CI/CD pipelines, IaC |
| Observability | Loki, Grafana, Prometheus, Elasticsearch, Kibana, ELK Stack |
Experience
Senior DevOps & AI Systems Engineer · SanMar
Architect and operate AI-augmented DevOps systems at enterprise scale, integrating LLM pipelines into daily operations while leading comprehensive CI/CD and infrastructure automation.
AI Systems & Automation
- Designed and built an end-to-end meeting intelligence pipeline: GPU-accelerated Whisper transcription → Claude API classification (13 categories) → structured summarization → automated action item extraction with Azure DevOps work item reconciliation.
- Developed AI-powered weekly activity reporting that scans ADO repositories for commits/PRs and generates executive narrative summaries via Claude API, replacing hours of manual status reporting.
- Authored 5 production mega-prompts with specification precision techniques: anti-drift safeguards, context-aware formatting rules, name/project correction mappings, and meeting-type-adaptive output structures.
- Built and maintain a 233-function PowerShell module library serving as composable context for AI-assisted development across 22 Claude Code projects with per-project memory tracking.
- Created 3 custom Claude Code skills for production use: ADO API resilience (retry/backoff/cascade prevention), PowerShell security auditing (6-category risk analysis), and Excalidraw diagram generation (27-point quality checklist).
- Developed custom DevOps tooling using OpenAI and Anthropic APIs for automated code analysis, documentation generation, and pipeline optimization.
DevOps & Infrastructure
- Led development of the "What's Deployed Where" system providing real-time deployment visibility across all environments, with automated notification systems and CAB weekly emails supporting nested ADO Work Item queries.
- Reverse-engineered Azure deployments to create custom IaC toolset (PowerShell/Azure CLI) for desktop and pipeline deployments; extended to PowerShell 7 and Linux for cross-platform flexibility.
- Redesigned serial deployment pipelines to parallel execution, achieving 48% deployment time savings; redesigned Azure deployments achieving 97% deployment time savings.
- Designed enterprise-wide Jira-to-Azure DevOps migration tooling around
jira-azuredevops-migrator— built automated notification systems, parallelized Linux repository updates, orphaned branch cleanup utilities, and environment tracking dashboards.
DevOps Engineer · Microsoft
- Designed CI/CD pipelines for Windows Photos App; analyzed workflows and implemented performance improvements.
- Designed automated test-failed-retry pipeline to isolate flaky infrastructure from genuine test failures — an early failure pattern recognition system.
- Implemented Docker-based ELK stack for build/release/test metrics collection and analysis.
- Built ADO activity dashboards and reporting using PowerShell and ADO REST API.
DevOps Lead · Garden City Group
- Led adoption of Git-based workflows and DevOps tooling for a new Azure-hosted product; trained teams on ADO, Git, GitFlow, and NuGet.
- Implemented CI/CD processes with dotCover, Karma/Istanbul coverage, and Elasticsearch/Kibana dashboards.
- Managed Agile workflow using Epics, Stories, and Work Items in ADO.
Senior Build Engineer · Intentional Software
- Created auto-deployment scripts connecting OpenStack/BlueBox cloud VMs to Jenkins in a Jenkins 1.6/2.0, SVN/Git, TFS environment.
Senior Build Engineer · FTI Consulting
- Administered Visual Studio TFS 2015; created automated code coverage metrics using dotCover and SQLCC.
- Co-designed division-wide dashboard reporting build status and coverage metrics via Elasticsearch and Kibana.
- Automated build monitoring, queuing, and team alerting using PowerShell, MSBuild, and TFS API.
Build Lead / Build Manager · Microsoft
- 10 years supporting enterprise-scale build/release infrastructure across Windows Media, Windows Mobile (100+ builds), Windows Live, Developer Division (Visual Studio), and console/services.
- Hired and managed a team of 5 build engineers providing 24/7 build and source management support.
- Designed multi-language Service Pack creation process, scaling from 4 to 26 languages.
- Built isolated build networks reducing build times and corporate network traffic.
- Created automated VM provisioning on Hyper-V, NuGet Gallery, and web-based build status dashboards.
- Implemented the Compatibility Labs program ("Windows Certified") for OEM hardware testing.
AI & Homelab Projects
TLDW (Too Long; Didn't Watch) — AI content intelligence system tracking 51 YouTube channels, 783+ videos with automated transcript processing, summarization, and detailed analysis. Active daily pipeline. More →
Transcripts Pipeline — Production Claude API pipeline: audio → Whisper transcription → Haiku classification → Sonnet summarization → action item extraction → ADO work item matching. Dual-model architecture with retry/backoff and state persistence. More →
Self-Hosted Infrastructure — 30+ Docker Compose services including AnythingLLM (document Q&A), n8n (workflow automation with scheduled AI integrations), Prometheus/Grafana monitoring, and media management. Complete infrastructure-as-code. More →
Kubernetes GitOps Cluster — Data-driven K8s infrastructure on Proxmox/Talos Linux with two-loop reconciliation: OpenTofu for infrastructure provisioning + Flux CD for application GitOps. MetalLB, Traefik, NFS storage, SOPS encryption.
Ansible Automation — Playbooks for system patching, reboot management, hardware inventory, and historical data collection across homelab fleet.