Rob Rogers

Senior Network Architect • Cybersecurity Developer • ISP Founder & Technical Lead • AI Systems Engineering

Patent US12494845B2 (Approved) Provisional Patent 64/105,371 (PCM)

Senior Network Architect, Cybersecurity Developer, and ISP Founder with over 20 years of hands-on systems engineering experience. Formerly reported directly to John McAfee as Lead Product Developer for the MGT Sentinel honeypot platform. Co-Founder & Vice President of Ontario High Speed Inc (~1,200 subscribers, 99.999% uptime), President of Legendary I.T. Solutions Inc, inventor of the approved Multi-Endpoint Radio Spectrum Redistributor (Patent US12494845B2), and co-inventor of the Pre-Generated Conversational Matrix (Provisional Patent 64/105,371) at mmaplabs.com.

My Projects

Troubleshooter - Origin Latest

Proof of concept for LLM-driven NPC AI in video games.

Troubleshooter: Origin is an immersive, Deus Ex-style game set in a dystopian world. Currently prototyped in Garry's Mod using Lua, it completely replaces traditional, rigid multiple-choice dialogue trees with dynamic Large Language Model (LLM) interactions.

Key Features:

Demo Video below for the "Barbara the Serial Killer" sidequest.
This is an older video that does not represent the current state of events. It will be updated at a later date.

Pre-Generated Conversational Matrix with Caching (PCM) Latest

Provisional Patent 64/105,371 • Developed at mmaplabs.com

The Pre-Generated Conversational Matrix (PCM) is a breakthrough data architecture engineered for on-device local AI and network-isolated, sovereign offline assistants. While originally benchmarked and battle-tested within complex, real-time video game NPC dialogue systems (where full-scale dynamic LLM generation proved computationally intensive and memory-bandwidth bound), PCM solves traditional edge inference barriers by shifting the paradigm from live token generation to intelligent semantic selection and kernel-level memory paging.

By pairing an offline high-parameter "teacher model" (e.g., 70B parameter models) with an on-device lightweight 1B or 2B classifier, PCM enables privacy-focused, zero-token-cost, instant offline conversational experiences on consumer hardware—without thermal throttling, battery drain, or extreme CPU and RAM bandwidth requirements. For complex dynamic queries, PCM features an intelligent cloud fallback and multi-LLM routing gateway that dynamically directs requests to optimal cloud models while caching responses to eliminate redundant API calls. Learn more at mmaplabs.com.

Core Use Cases:

Provisional Patent Architecture & Innovations:

Target Architecture & Engine Integration

Architected to support cross-platform native execution, embedded devices, Unreal Engine, and Source Engine environments through a clean, decoupled pipeline.

Wi-Fi Device Tracking – E-Tagged (2015)

This was a proof of concept unit made in a weekend over beers with my business partner out of a previous project. It was for tracking 802.11 devices associated (Connected to wifi networks) and unassociated (wifi enabled on but not connected to anything).

Initially built in a more limited prototype form in 2012 using off-the-shelf wireless gear, the project hit severe hardware constraints with insufficient RAM and flash memory to support heavy packet processing and local metric aggregation. Recognizing these physical limits, we transitioned the architecture to dedicated, custom embedded hardware designed specifically for high-throughput wireless monitoring.

Later versions had a complete web portal, proper database structure, google maps integration, promiscuous bluetooth and discoverable bluetooth, as well as tracking access-points after we started noticing more cars with in-car wifi built in. I also made a version in 2019 that could also track tire pressure monitoring sensors' unique 16-bit ID numbers, associating all metrics it discovered together. When paired with a camera you could build a pretty expensive inventory of metrics to track people or cars.

Fixed Wireless LTE Prototype

Fixed wireless LTE unit for long-range LTE connections in remote areas.

Field prototype engineered for clients with poor cellular connectivity, specifically designed for use in elevator machine rooms where it served as the sole reliable method to establish signal. Demonstrates signal distribution, RF link budgeting, and wireless broadband access node architecture for high-speed connectivity in rural and underserved areas.

Sentinel (2016–2020)

Deception Honeypot System • Lead Product Developer & Architect

Sentinel was a high-interaction deception and intrusion detection system ("honeypot in a box") designed to detect unauthorized network probing, track attacker dwell time, and alert enterprise administrators to threats in real time. Operating on behalf of Ontario High Speed Inc, I reported directly to John McAfee and worked closely under his leadership in Durham, North Carolina as Lead Product Developer and Principal Architect.

Together with John McAfee and security researcher Chris Roberts, I architected the complete platform, led a ground-up rewrite to meet strict enterprise auditing standards, built the hypervisor/WAF server infrastructure, authored the core Python/C++/C embedded codebase, and managed the software engineering team. See detailed technical experience breakdown below under Professional Experience.

Click images to view full size.

Elevator Monitoring System – Liftronics (2022)

Designed and deployed an end-to-end remote telemetry and diagnostic platform engineered for real-time elevator controller monitoring. Starting with the Otis elevator controller, we developed a modern drop-in replacement for the legacy Otis REM system (which historically relied on manual DTMF phone audio tones to report status).

This system features a cloud-based management portal that proactively identifies fault codes before service disruptions occur, enabling preventative maintenance and preventing passenger entrapments. In the event of a critical alert, technicians receive instant SMS and dashboard telemetry detailing exact fault codes, ensuring proper replacement components are brought directly to the site for rapid restoration.

Hardware & Architecture Highlights:

Click images to view full size.


Professional Experience

2018–Present — Legendary I.T. Solutions I.N.C.

President & Co-Founder | Office: 78 Centre St N, Floor 2, Unit D, Oshawa, ON | legendaryitsolutions.ca

2010–2017 — Ontario High Speed Inc

Vice President / Co-Founder (ISP Operations)

ISP Operations & Engineering (Vice President / Co-Founder)

Founded to deliver high-performance rural broadband, Ontario High Speed built out dedicated wireless distribution towers and a fiber-backed carrier network serving subscribers across the Durham region before successfully selling the customer base to Vianet.

Click images to view full size.

Enterprise IT Consulting (Durham Region)

MGT Capital Investments, Inc. / Sentinel Project Development (May 2016 – Nov 2017)

Lead Product Developer & Principal Architect — Reporting Directly to John McAfee (Durham, NC)

Click images to view full size.

2009–2010 — Silonet.ca (Orono, ON)

Network Engineer

2006–2009 — BestByte Computers (Oshawa, ON)

Part-Time Service Manager (College Job)

2004–2006 — Canampak Computers

Computer Repair Technician & Sales (High School Part-Time Job)

2003–2004 — Global Electronics Systems (GES)

Computer Technician & Technical Sales (High School Part-Time Job)


My Patents

Patent US12494845B2 (Approved December 2025) — Multi-Endpoint Radio Spectrum Redistributor

System for extending cell phone service, FM radio, UHF/VHF television, and public safety radio (police/fire/EMS) signals from urban areas into geographically isolated regions over fiber-optic transport using optical splitters to support full bidirectional communication. Prototypes successfully tested at PARI Radio Telescope facilities.

Prototype hardware board for the approved Multi-Endpoint Radio Spectrum Redistributor patent

Provisional Patent 64/105,371 — Pre-Generated Conversational Matrix with Caching (PCM)

Developed at: mmaplabs.comCo-Inventors: Robert Rogers & Adrian Chadd

Read-optimized data architecture and semantic caching system enabling lightweight 1B/2B parameter models to deliver on-device local AI, sovereign offline assistants, and enterprise upstream query caching. Bypasses live token generation bottlenecks ("Inference Wall") using sub-5ms term-frequency pre-filtering, 1B/2B classification gateways, virtual memory OS kernel page-faulting (`mmap`), an ephemeral Context Vault state machine, and intelligent cloud fallback with dynamic multi-LLM routing.


Books, Social Media & YouTube

My Book

The John McAfee Experience Book on Amazon

My Facebook Public Page

Official Public Facebook Page

Rob Rogers Official Facebook Profile

My GitHub

Sometimes I make open-source projects and post them here; not all are complete. Most of what I do is proprietary and closed source.

https://github.com/Rob-Rogers

My YouTube Channel

Rob Rogers @robsyoutube YouTube Channel

Rob Rogers YouTube Channel

Education & Formal Training

Durham College Oshawa Campus — Computer Systems Technology (3-Year Advanced Degree) | Class of 2009

★ Featured Alumni Spotlight
Featured Durham College Notable Alumni Article: Partnering with McAfee Global Technologies

Comprehensive 3-year enterprise IT infrastructure, systems administration, and software engineering curriculum covering on-premises enterprise environments, network architecture, and security engineering:

Certifications: CompTIA A+ (Certified 2009)


Get In Touch

Send a direct message regarding projects, consulting, or inquiries.