LIMITED ENERGY EXPERTS
AI does not replace the design engineer, just as the tractor did not replace the farmer. It is a tool. A well-qualified and experienced engineer of record is required to drive the AI tractor.
Shawn C. Tovey, RCDD
I design physical security and ICT systems for environments where failure carries real consequence: federal installations, hyperscale data centers, airports, hospitals, correctional facilities, and enterprise campuses. The work is about anticipating how systems fail and designing to prevent it. The RCDD credential from BICSI is the formal acknowledgment of that discipline.
The tools and projects on this site exist because good engineering problems deserve better instruments. The free tools are available to anyone in the limited energy community. The research is shared because the field improves when practitioners understand the vulnerabilities they are designing around.
Over two decades designing and supporting integrated technology systems across a wide range of environments, from single facilities to large multi-site deployments spanning Division 27 and Division 28 systems.
Experience spans education, healthcare, aviation, government, manufacturing, enterprise campuses, correctional facilities, and hyperscale data centers. Work ranges from individual facilities to nationwide deployments, including design development, construction documents, design-assist delivery, proposal development, and owner-stakeholder coordination. Volume of proposal activity has reached nine figures across physical security architecture alone.
The list below reflects manufacturer training and platform exposure carried through years of design and field work, not a dated certification roster. The work shows what the training built.
Each project started as a problem I needed to solve. They are engineering explorations: built to work, shared to be useful, and continued as long as they remain interesting.
Why: Telecom room power coordination was a repeated pain point: sizing UPS, generators, PoE budgets, and cooling loads by hand from scattered references on every project.
What it solves: A single browser-native calculator that handles UPS sizing, generator sizing, PoE budget, and cooling demand for telecom rooms and equipment rooms. No login, no server, no data leaves the browser.
Technologies: Vanilla HTML, CSS, JavaScript. Runs entirely client-side. Aligned to BICSI and TIA standards.
Where it's headed: Continued refinement based on field use and community feedback. The goal is to reduce the class of power coordination errors that happen when engineers estimate UPS and generator requirements from memory or incomplete references.
Launch PLEX →Why: Limited energy estimating still relies heavily on general-purpose spreadsheets that were never designed for Division 27/28 scope or BICSI work breakdown structures.
What it solves: A complete job estimator built around BICSI WBS codes. Configurable crew roles, workweek patterns, and federal or California overtime rules. Generates a customer-ready proposal and an internal bid report with financial roll-up, bid health indicators, and risk flags. No account required.
Technologies: Vanilla HTML, CSS, JavaScript. Entirely client-side. No data transmitted.
Where it's headed: Expanding scope coverage and improving risk flag logic: the bid health indicators that surface problems before a job is priced wrong and the damage is done downstream.
Launch The Estimator →Why: Local networks accumulate unknown devices, exposed ports, outdated services, and unnoticed configuration drift. Most home and small office environments have no repeatable way to see what's actually running on the network.
What it solves: A local-first network awareness and vulnerability assessment tool that runs automated subnet audits (ARP device discovery, port scanning, and CVE assessment via nmap scripts) and streams results live to a browser dashboard. Generates a PDF report with findings, risk level, CVSS scores, and recommendations. Tracks device baseline changes between scans.
Technologies: Python, Flask, Flask-Sock, HTML, Shell, nmap, arp-scan, WeasyPrint, CVE/CVSS lookup. Runs on Linux/Kali Purple.
Where it's headed: Improving baseline comparison, remediation guidance, and reporting so small environments can identify weak points and track progress before those weak points become operational or security problems.
Why: Modern AI assistants each have unique strengths and limitations. VolTex coordinates a structured volley of analysis between independent AI agents to produce recommendations built on consensus, not a single perspective. You make the final call.
What it solves: A local-first multi-agent AI orchestration platform. Independent agents bring different strengths and perspectives to the same problem. Ideas are exchanged, challenged, and refined through a bounded process. One designated execution agent interacts with approved tools. The others focus on independent reasoning and review. The consensus recommendation is presented to you for approval.
Technologies: Multi-agent orchestration, Git Integration, Desktop Commander, Chrome MCP, Discord operator interface, structured consensus protocol, local inference.
Where it's headed: Expanding into a general-purpose collaboration platform for engineering reviews, technical writing, documentation, planning, research, and design critiques. Any domain where independent AI perspectives improve decision quality while preserving human oversight.
Learn more →Why: Cloud-dependent home automation creates single points of failure, data exposure risks, and subscription lock-in. Local inference enables private AI tooling without sending sensitive information to external services.
What it solves: A dedicated home server running a private inference stack and home automation platform. All private engineering tools run here. Sensor network covers environmental monitoring, energy monitoring, lighting, appliances, and access. Dashboards built in Grafana against InfluxDB time-series data.
Technologies: Docker, Ollama, Open-WebUI, Home Assistant, Node.js, Grafana, InfluxDB, Tailscale, YoLink, Tesla Powerwall, TP-Link Kasa.
Where it's headed: Deeper integration between the inference stack and automation triggers, and improved energy visibility through solar and Powerwall data, building toward a home that maintains capability when normal infrastructure degrades.
See the build →Not a credentials list. A window into what I find genuinely interesting, and where my thinking is going.
The thinking behind the tooling.
I view AI as an engineering accelerator, not a replacement for expertise. My workflow combines industry standards, practical field experience, structured peer review, and AI-assisted automation to improve consistency, documentation quality, and engineering efficiency, while maintaining human oversight at every decision point that matters.
Engineering is fundamentally about reducing uncertainty. Every standard, every calculation, every review cycle exists to narrow the gap between what was designed and what gets built, and between what was intended and what fails. AI is one tool in that effort. Engineering judgment remains the most important component. Technology assists; engineers remain accountable.
The RCDD drives. The tools handle the repetitive coordination work. A model that understands TIA-568 and can cross-reference BICSI TDMM against a project specification is a force multiplier. It still requires the credentialed practitioner who must review it, own it, and stand behind it. That line does not move.
Everything sensitive runs on dedicated on-premise hardware. The stack is built on open-weight models grounded in real standards language, not generalist models that hallucinate clause numbers. Secure remote access via Tailscale keeps it off the public internet entirely.
Overviews of the tooling stack: the design philosophy, how PLEX and The Estimator work, and why these tools were built for the limited energy trade.