Twenty years on
one question.
How do you deploy a system that decides things on a person's behalf, somewhere being wrong is expensive, and get people to trust it enough to actually use it? I've worked on it in eight settings, from a special operations exoskeleton to a 911 call center to an institutional AI platform. PhD in Systems Engineering.
I want a bigger platform to run AI on. Not a bigger title.
Senior, Lead or Principal product and strategy roles, somewhere with the scale to build AI systems that hold up. Track record's below.
View Full Work →Redfearn Group takes the engagement your team is stuck on.
AI strategy. LLM agent architecture. Governance frameworks that satisfy legal without slowing engineering down.
Email brady@redfearn.group →One question, eight settings.
The domain changed each time. The question did not.
MS thesis, BYU
What an organization pays when it adopts an unfamiliar practice. The enterprise AI adoption problem, before the AI.
PhD dissertation, Virginia Tech
What an intelligent system should surface to an operator under load, when it should defer to the human, and how it should fail safely.
TALOS, Special Operations Command
The human-AI interaction model for a powered exoskeleton, worked out on site with operators under hard safety constraints.
Advanced HCI, BYU
A curriculum built around adaptive interfaces: how people direct, correct, and override systems that decide for them.
Army Cyber Institute, West Point
Applied LLM research on adversarial AI risk in critical infrastructure, and two published assessments.
Next Generation 9‑1‑1
Whether AI can be trusted inside a 911 call center, where a wrong output costs a life.
Institutional AI platform
The product spec for an enterprise AI platform serving more than 200,000 learners.
Title 15 analysis, Spanish Fork
Publishing the weights: an independent public analysis with every calculation shown, so anyone can check whether the conclusion survives.
Track Record
Where being wrong is expensive.
Defining what an exoskeleton decides, before the models could run it
What the system surfaces to an operator under load, when it defers to the human, and how it fails safely. Secured $750K in research funding along the way.
Deciding where AI belongs in a 911 call center
A wrong model output there costs a life. I built the governance criteria for making that call, and the vendor assessments behind it drove pilot and procurement decisions.
Holding six contracts worth $45M to one set of delivery standards
An org of 12 across five teams, working to the delivery standards I set.
Choosing an AI platform from 20+ vendors on weighted evidence
Benchmark data, published weights, and a recommendation senior leadership used to set platform selection and a six-capability roadmap.
Case studies, not slideware.
Scaling to 4 Concurrent AI Programs and Architecting the Platform to Unify Them
Took an institution from one AI initiative to four running in parallel, then built the platform architecture that let them share one foundation.
Deciding Where AI Can Be Trusted in a 911 Call Center
Governance criteria for where AI can be safely deployed in 9-1-1, where a wrong output costs a life, and the assessments that turned them into pilot and procurement decisions.
Managing Six Contracts Worth $45M
Six contracts worth $45M, individual contributor work on 14 programs, an org of 12 across five teams, a 0-to-1 procurement platform, and applied LLM research on adversarial AI risk at the Army Cyber Institute.
Capabilities
AI Product Strategy
Roadmaps that move adoption and revenue, not just ship features.
LLM Agent Architecture
Multi-agent systems designed for reliability, not demos.
AI Governance & Compliance
NIST-aligned frameworks that enable shipping instead of blocking it.
AI-Native Team Leadership
Built 7 purpose-built AI skills that orchestrate retrospectives, documentation, and research synthesis across a team's active workstreams.
Systems Engineering
Take a complex system apart, work out how the parts interact, and improve one component at a time.
Enterprise & Regulated Environments
Security and compliance-heavy organizations, by choice. Higher education, healthcare, and large enterprise, where releases have to clear legal and security review.