PATENT
Patents & Intellectual Property

Invention
as evidence.
Not résumé.

These patents were not drafted in isolation; they came out of production platform work by cross-functional engineering teams. Frank is a named inventor on 12 filings across 4 patent families — arising from designing and delivering real systems at enterprise scale.

Each filing documents a solved engineering problem that met USPTO novelty standards. The surrounding platforms went on to serve large-scale, real-world workloads, which is why this body of work matters for technical and board-level decision making.

4
Issued patents
12
Total filings
4
Patent families
2023
Most recent filing
The origin of this IP

Between 2014 and 2023, Frank Moss served as founding architect of Lumen Technologies’ global edge cloud platform — a system designed to deploy environment-tolerant compute at every node along the network, from last-mile connections through the global internet backbone. The goal was sub-5ms response times for latency-sensitive applications: industrial robotics, 5G edge workloads, real-time automation.

That problem required breakthrough thinking in three areas: how hardware could be dynamically composed and decomposed across a distributed network (composable infrastructure), how traffic could be routed with network-level intelligence rather than static rules (intent-based orchestration), and how AI inference could be delivered as a managed service at the edge. The patents below document the solutions.

All listed patents were assigned to Level 3 Communications / CenturyLink / Lumen Technologies, where the work was performed. Frank Moss is a named inventor alongside team contributors. The IP ownership remains with the assignees; inventorship reflects technical contribution, not personal ownership.

Patent families · click any to expand filings
Family 01

Disaggregated & Distributed Composable Infrastructure

A system and method for dynamically composing and decomposing physical compute infrastructure across distributed network nodes — enabling hardware resources to be assembled on-demand for latency-sensitive workloads and released when no longer needed. The foundational patent behind the <5ms edge cloud platform.
Expand filings ↓
Issued
3 filings
Issued
Disaggregated & Distributed Composable Infrastructure
USPTO No. 11,425,224 · App. 17/891,775
Issued Aug. 2022
Pending
Disaggregated & Distributed Composable Infrastructure (U1)
App. No. 17/184,879
Filed Feb. 2021
Pending
Disaggregated & Distributed Composable Infrastructure (C1)
App. No. 17/891,775
Filed Aug. 2022
Family 02

Intent-Based Orchestration Using Network Parsimony Trees

A method for routing and orchestrating network traffic using intent-based logic and parsimony tree structures — enabling the network to determine the most efficient path for traffic based on declared intent rather than static routing rules. Three continuation patents have issued from this core invention.
Expand filings ↓
Issued ×3
5 filings
Issued
Intent-Based Orchestration Using Network Parsimony Trees
USPTO No. 11,343,201 · App. 17/176,062
Issued May 2022
Issued
Intent-Based Orchestration Using Network Parsimony Trees (C1)
USPTO No. 11,509,601
Issued Nov. 2022
Issued
Intent-Based Orchestration Using Network Parsimony Trees (C2)
USPTO No. 11,637,790
Issued Apr. 2023
Pending
Intent-Based Multi-Tiered Orchestration and Automation (U1)
App. No. 17/176,062
Filed Feb. 2021
Pending
Intent-Based Orchestration Using Network Parsimony Trees (C3)
App. No. 18/137,791
Filed Apr. 2023
Family 03

Block-Level, Bit-Mapped Binary Data Access for Parallel Processing

A method for accessing and processing binary data at the block level using bit-mapped structures to enable high-performance parallel processing. Addresses the fundamental throughput bottlenecks in distributed data systems operating at edge scale.
Expand filings ↓
Provisional
2 filings
Provisional
Block-Level, Bit-Mapped Binary Data Access for Parallel Processing
USPTO No. 63/511,378
Filed Jun. 2023
Pending (U1)
Block-Level, Bit-Mapped Binary Data Access for Parallel Processing (U1)
1759-US-U1
Signed 2024
Family 04

Inference as a Service

A system for delivering AI model inference as a managed network service — enabling inference workloads to be run at the point in the network closest to the data source, rather than centralized in a cloud data center. Filed in September 2023, this invention directly addresses the infrastructure challenge at the center of the current AI deployment era: how to run inference at scale, at the edge, with acceptable latency and cost.
Expand filings ↓
Provisional
2 filings
Provisional
Inference as a Service
USPTO No. 63/581,842
Filed Sep. 2023
Pending (U1)
Inference as a Service (U1)
1775-US-U1
Signed 2024
Why this IP matters for companies today
Inference as a Service is the AI infrastructure problem of 2024–2026.
Every company deploying LLMs at scale is grappling with the same question: where do you run inference, and how? Centralized GPU clusters are expensive and introduce latency. Edge inference solves both problems. This patent — filed before most companies had articulated the problem — documents a specific architectural solution.
Composable infrastructure is the architecture model for AI hardware flexibility.
The AI era has made static hardware allocation untenable. Training, fine-tuning, inference, and retrieval have radically different hardware profiles. The composable infrastructure patent family describes systems for dynamically assembling and disassembling hardware configurations — an approach now central to how hyperscalers think about resource management.
Intent-based orchestration is how modern networks handle AI workload routing.
As AI workloads proliferate, routing them efficiently across hybrid infrastructure requires logic that goes beyond static rules. The network parsimony tree approach describes a mathematically principled method for routing based on declared intent — a concept that maps directly to how current network intelligence platforms are designed.
The filing dates matter as much as the patents themselves.
Priority dates establish what was known and solved when. The composable infrastructure and orchestration patents were filed in 2021, the inference patent in 2023. These predate most of the commercial AI infrastructure conversation. That earliness is evidence of original thinking, not implementation of what others had already articulated.

All patents are publicly searchable through the USPTO Patent Full-Text Database and Google Patents. The issued patent numbers — 11,425,224 · 11,343,201 · 11,509,601 · 11,637,790 — can be searched directly. Provisional applications are not publicly published until their utility application counterparts are published; the application numbers are provided for reference.

Frank does not offer licensing for these patents because he is not the owner. He is available for architecture strategy, invention process guidance, and technical advisory informed by this work. Contact frank@ignition-sequence.com.

USPTO 11,425,224 ↗ USPTO 11,343,201 ↗ USPTO 11,509,601 ↗ USPTO 11,637,790 ↗

If these problems are yours to solve, let’s talk.

Advisory engagements, technical collaboration, and strategic counsel for companies where AI infrastructure, edge compute, network intelligence, or defense and SLED architecture are at the center of the product strategy.

Advisory inquiries View engagement structures Direct contact