Senior Automation Engineer building AI + DevOps systems, local LLM infrastructure, MCP workflows, and focused utility products.
I turn operational friction into software: self-hosted AI control planes, automation platforms, local inference stacks, and small apps that solve specific real-world problems.
I work at the intersection of automation engineering, AI infrastructure, DevOps platforms, and small utility products.
- AI agent infrastructure — memory layers, MCP servers, prompt/skill systems, local-agent workflows
- Local/private LLM stacks — OpenAI-compatible APIs, OpenVINO, Windows AI hardware, GPU/NPU experimentation
- DevOps automation platforms — Terraform, Ansible, Kubernetes, CI/CD, observability, internal tooling
- Operator-first apps — focused Android / Wear OS tools with privacy-first architecture and Play Store readiness
- Consulting/product systems — portfolio, technical content, and practical AI infrastructure offers
Build the tool once. Remove the task forever.
These public repos best represent my current direction.
| Project | What it is | Stack / Focus | Status |
|---|---|---|---|
| MemoryOps | Self-hosted memory control plane for AI agents | Rust, Postgres, Redis, Qdrant, MCP, control UI | Alpha |
| OpenVINO Windows LLM | Local OpenAI-compatible LLM server for Intel Windows PCs | Python, OpenVINO GenAI, Windows, CPU/GPU/NPU, web UI | Working |
| AgentDefaults | Reusable defaults for AI agents, skills, prompts, wrappers, and token-efficiency workflows | Markdown, validation scripts, AI agent UX | Active |
| npu-windows | Earlier Intel NPU local LLM experiment and predecessor to the OpenVINO stack | Windows, Intel NPU, local AI | Legacy / reference |
MemoryOps is my self-hosted memory layer for AI agents. It is designed for teams that need governed, inspectable, token-aware context retrieval across engineering activity instead of one-off prompt stuffing or unmanaged vector search.
Focus areas:
- Structured ingestion from engineering systems and agent observations
- Episodic → semantic memory lifecycle management
- Hybrid retrieval, feedback, decay, pinning, pruning, and auditability
- MCP/API access so coding agents can retrieve and store memory directly
- Operator UI for visibility into what agents remember and why
OpenVINO Windows LLM turns Intel Windows PCs into practical local AI workstations. It wraps OpenVINO GenAI in an OpenAI-compatible server with streaming chat, model lifecycle management, a built-in browser UI, conversion helpers, and direct targeting for Intel CPU, GPU, NPU, and AUTO devices.
Use cases:
- Open WebUI / n8n / LangChain / Continue integration
- Local-first AI experimentation without cloud APIs
- Intel NPU/GPU/CPU device routing tests
- Laptop-friendly model conversion and serving workflows
AgentDefaults is a practical library of reusable agent profiles, skills, prompts, wrappers, examples, and validation patterns. It is built around a simple idea: AI engineering improves when default behavior is reusable, testable, and benchmarkable.
It includes:
- Repo-aware coding-agent defaults
- Claude / Gemini / Copilot / editor wrappers
- Token-efficiency skills and measurement prompts
- Reusable task profiles for DevOps, research, SEO, coding, and documentation
- Validation scripts for keeping prompt libraries maintainable
Not everything I build is public. A lot of my most active work lives in private repos because it includes product-stage code, app-store release details, security-sensitive automation, or client-style infrastructure patterns.
Current private/product areas include:
| Area | What it demonstrates |
|---|---|
| Wear OS utility apps | Kotlin, Jetpack Compose, Play Billing, tiles, complications, haptics, Android release engineering |
| WebhookDeck | Secure wrist-based webhook triggers for operators, homelab users, and automation workflows |
| WristSense / WristNote / WristDash / WristLux | Sensor dashboards, voice notes, Home Assistant controls, light metering, phone-watch sync |
| FidgetDrop | Privacy-first Android haptic product with one-time Pro unlock and local-only state |
| InfraOrbit | Local-first AI command center for DevOps and platform engineering teams |
| Consulting site | AI infrastructure consulting hub, technical articles, videos, app pages, legal/privacy docs |
The public repos are the visible slice. The private repos are where the productization, release hardening, and app-store workflows happen.
| Layer | Tools |
|---|---|
| Cloud & Infra | Azure, AWS, OCI, Terraform, Ansible, Helm |
| Containers & Platforms | Docker, Kubernetes, AKS, GitHub Actions, Azure DevOps, Jenkins |
| Languages | PowerShell, Python, Go, Rust, TypeScript, Kotlin, Bash |
| AI / Agents | MCP, FastMCP, local LLMs, OpenVINO, Open WebUI, n8n, Codex, Claude, Gemini, Perplexity |
| Backends | Django REST Framework, FastAPI, Go HTTP services, Postgres, Redis, Qdrant |
| Frontends | React, Vite, TypeScript, Flutter, Jetpack Compose, Compose for Wear OS |
| Observability | Grafana, Prometheus, Sentry, structured logging, health checks |
| Mobile / Wearables | Android, Wear OS, Play Billing, Data Layer sync, tiles, complications, haptics |
- Building self-hosted memory and context systems for AI agents
- Making infrastructure more agent-accessible through MCP, APIs, and typed control surfaces
- Running and evaluating local/private AI on Windows, GPU, and NPU hardware
- Turning DevOps patterns into productized internal platforms
- Shipping narrow, privacy-first utility apps instead of vague demos
- Creating technical content around AI infrastructure, automation, and local LLM workflows
- 🌐 Portfolio: profile.quinnfavo.com
- 🧠 Consulting: consultant.quinnfavo.com
- 💼 LinkedIn: linkedin.com/in/quinnfavo
▶️ YouTube: youtube.com/@QuinnFavo
Automation over ceremony. Utility over novelty. Systems that compound.