Infrastructure automation and network security engineering for AI systems
Zero-trust architecture · local-LLM reliability · Python automation · RTP / Triangle, NC
Website · LinkedIn · Writing · Video · stew@stewalexander.com · Book a call
I design and automate secure infrastructure for AI and ML workloads — zero-trust networks, multi-cloud platforms, and the low-latency fabrics that robotics and inference systems depend on. Most of my current open-source work sits at the point where that discipline meets local language models: making them private, durable, and honest.
Current focus — Aida, an offline memory and integrity runtime for models running on Ollama, and CERBERUS, a multi-perspective framework for quantum-era cyber defense.
Open to — consulting, fractional SRE, architecture review, speaking, and collaborators on the work below.
Active work
| Project | Status | Links |
|---|---|---|
Aida (continuous-ai) — offline memory and integrity runtime for local LLMs |
Active development; cross-platform, MLX and LanceDB | Repository · Site |
| CERBERUS — multi-perspective AI framework for quantum-era defense | Architecture and writeups; seeking collaborators | Overview |
| AI–Human Mutualism — guardrails for higher-signal human/model exchange | Specification published | Repository |
| ShadowVendor — MAC and vendor analysis with SIEM export | Maintained; contributions welcome | Repository |
| Live Tech News — self-refreshing news board built on GitHub Actions | Shipping | Repository |
Currently learning post-quantum cryptography implementation (ML-KEM, ML-DSA) and Rust for performance-sensitive network tooling.
AI systems and model reliability
| Project | Description | Stack | Stars |
|---|---|---|---|
| continuous-ai | Fully offline memory and integrity runtime for Ollama models. Memory you own, audit, and correct. | Python, MLX, LanceDB |  |
| python-tutor | Offline-first Python tutor with a local model, sandboxed runner, and source-backed feedback. | Python, FastAPI, Gemma |  |
| AI-Human-Mutualism | Guardrails intended to leave both parties in a conversation clearer than they entered. | Specification |  |
Network and security tooling
| Project | Description | Stack | Stars |
|---|---|---|---|
| ShadowVendor | Turns MAC address tables into vendor dashboards, flags new devices, and exports SIEM events. Cisco, Juniper, Aruba, Extreme, Brocade. | Python |  |
| network-attack-visualizer | Step-through walkthroughs of Layer 2 and 3 attacks with packet flow, topology, and ATT&CK context. Demo | D3.js, no backend |  |
| Linux-Tools | The toolkit I install on every Debian host, with a single-command installer for all of it. | Python, TUI |  |
| maclookup.py | Offline OUI and vendor lookup accepting any MAC address format. | Python |  |
| MAC2CSV | Converts Cisco show mac address and show ip arp output to spreadsheet-ready CSV. |
Python |  |
References and learning tools
| Project | Description | Stars |
|---|---|---|
| VIM-Awesome-Cheatsheet | A dense, single-page Vim reference. |  |
| Awesome-Mnemonics | Field guide to mnemonics for engineers: outages, root cause analysis, on-call, and burnout. |  |
| Python-Power-User | Forty-seven section Python reference with an interactive curses interface. |  |
| VSCODE-mobile-Cheat-Sheet | Every VS Code shortcut, mobile-first and offline capable. Demo |  |
| NetDisco-Ubuntu-Install | Clearer installation instructions for NetDisco on Ubuntu. |  |
Web applications and experiments
| Project | Description |
|---|---|
| live-tech-news | Rotating live tech news board with fuzzy search and a ninety-day archive, refreshed by GitHub Actions. |
| MebaneWeather | Severe weather dashboard drawing on NWS, SPC, and Open-Meteo. Live |
| global-tensor-network | Interactive tensor model of fifteen coupled global systems with documented couplings. |
| Fractal-Mandala-3D | Seven layers of sacred geometry rendered as one Three.js journey inward. Demo |
| rain-view | Contemplative ambient rain with generative piano. Demo |
| tropical-todo-list | Local-first, offline-capable task list. Demo |
Technical depth
| Domain | Detail |
|---|---|
| Linux and UNIX | Debian and RHEL, system hardening, fleet automation at scale |
| Networking | Multi-vendor Cisco, Juniper, Dell, and Extreme; BGP, SD-WAN, high-availability wireless, sub-millisecond fabrics for AI and robotics |
| Security | Zero trust to NIST 800-207, SIEM engineering, next-generation firewalls (SRX, Palo Alto, FortiGate), PCI-DSS gap analysis |
| Cloud and infrastructure as code | AWS at two hundred-plus instances, GCP, Azure, Terraform, Kubernetes, OpenStack |
| Automation | Python, Ansible, SaltStack, CI/CD pipelines |
| AI systems | Local model deployment, multi-agent architecture, confabulation reduction, zero-trust AI workloads |
| Post-quantum cryptography | ML-KEM and ML-DSA implementation readiness |
Engagements and availability
| Client type | Engagement |
|---|---|
| Remote teams and startups | Zero-trust architecture, multi-cloud design across AWS, GCP, and Azure, low-latency networking for AI and robotics |
| Small and mid-size businesses in the Triangle | Fixed-fee security assessments, PCI-DSS gap analysis, ransomware readiness and hardening |
| Managed service providers | White-label security architecture and automation capacity |
| Conferences and podcasts | Talks on quantum security, AI infrastructure, and zero-trust design; workshop facilitation |
Background and recognition
| Organization | Work |
|---|---|
| Covariant.AI | Hybrid-cloud infrastructure for AI robotics; Python and SaltStack automation to shorten deployment cycles |
| Perrone Robotics and IIHS | Autonomous vehicle networks supporting crash-test programs |
| US EPA | Government cloud implementations |
| Alamance Community College | Educational network infrastructure |
| IBM, Lucent Technologies, Extreme Networks | Enterprise networking and support engineering |
Technical Superstar Award, Extreme Networks (2009) · Who's Who in America honored listee (2024) · Upstream contributions to LibreNMS, OSQuery, and NTOPng
Writing and video
| Channel | Subject |
|---|---|
| Medium | Command-line tooling, infrastructure automation, and quantum security. Most-read piece: 20 Awesome Command Line Tools for the Mac. |
| Cerberus-AI on YouTube | Quantum cybersecurity explainers and CERBERUS demonstrations |
| Quantum Shield Initiative | Post-quantum and AI security research with implementation guidance |
Contributing
| Repository | Useful contributions |
|---|---|
| continuous-ai | Retrieval quality, platform packaging, evaluation harness |
| ShadowVendor | Dashboard features, vendor fingerprints, additional SIEM formats |
| network-attack-visualizer | New attack scenarios and ATT&CK mappings |
| Awesome-Mnemonics | New entries with guidance on when to use them and their failure modes |
Please open an issue before starting substantial work. Collaborators are welcome on CERBERUS.