Designing What an Exoskeleton Decides, Before the Models Could Run It

Situation

TALOS, the Tactical Assault Light Operator Suit, was U.S. Special Operations Command's powered exoskeleton program. A system worn by an operator makes decisions on that person's behalf, in the field, where a wrong one is dangerous. The open questions were what it should decide, what it should show the operator, and when it should hand control back. The models that could run a system like that did not exist yet. This ran concurrently with my PhD.

What I Built

• Defined the human-AI interaction model for the system: what it decides, what it surfaces to an operator under load, when it defers to the human, and how it fails safely.
• Led product design for the head-worn mixed reality interfaces, working on site with operators under hard performance and safety constraints.
• Secured $750K in competitive research funding by designing the business cases and product strategy behind two competitively awarded R&D programs.
• Launched two prototypes on time and under budget, saving $100K while meeting compliance and client standards.
• Ran market and competitive analysis that set roadmap priorities.

The Hardest Problem

Designing the interaction before the technology could run it. There was no model to test against, so the rules for what the system surfaces, when it defers, and how it fails had to be worked out with operators on prototypes, then held to hard performance and safety constraints.

Outcome

$750K in competitive research funding across two competitively awarded R&D programs, two prototypes shipped on time and $100K under budget, and an interaction model that preceded the models capable of running it. The same four questions sit under every AI deployment I have worked on since.

Skills Demonstrated

Human-AI interaction design, product strategy and business cases for research funding, field research with operators, delivery under safety and compliance constraints.