Selected Prior Work
Case studies from Amazon, NASA, and Mercedes-Benz.

Devices Design Group
Human Factors Researcher
Mapping Cognitive Load to Echo Device Design
Latency and Human Factors work shaped Echo Show and Echo Hub experiences affecting 75M+ active customers.
I co-established Amazon's first neuroimaging framework — combining fNIRS brain scanning with eye-tracking to quantify cognitive load across streaming UIs, replacing assumption-driven design with direct neural and gaze evidence.
Neuroimaging Methodology
Applied a controlled within-subjects fNIRS protocol measuring prefrontal cortex activation, synchronized with eye-tracking to separate gaze behavior from cognitive effort. Findings translated directly into product requirements, shaping the interaction design of Echo Show and Echo Hub — reaching 75M+ active customers.
Context-Aware Interaction Patent
Beyond neuroimaging, I co-invented a context-aware interaction system — where the same physical input routes to contextually appropriate actions based on active UI state — granted as US Patent 12,532,040 B1 in January 2026. This work earned the 2023 Amazon Inventor Award.
Calibrating Alexa Response Latency to Human Perceptual Limits
I established Alexa's latency satisfaction thresholds, then Alexa Economics projected $50M in incremental operational value over 15 months if those latency targets were met across high-priority voice intents.
Alexa's response latency targets were defined by engineering constraints, not human perception. I led a multi-year Human Factors research program to establish empirically grounded thresholds — identifying precisely where response delay crosses from imperceptible to trust-eroding, and anchoring those findings in the economic models that drive product investment.
Isolating the Latency Satisfaction Threshold
Used a Wizard of Oz latency injection system to manipulate visual and auditory response latency on users' devices across thousands of trials — controlling latency as an isolated variable while keeping all other system factors constant. Analysis of the resulting data identified a High Satisfaction Threshold (HST) at 1,000ms: the point where user tolerance breaks down, "too slow" ratings climb sharply, and satisfaction drops in a consistent, reliable pattern.
NVA Model Integration & Economic Validation
The HST was formally adopted by Alexa Economics and embedded into the Negative Value Action (NVA) model — redefining any response exceeding 1,000ms as a quantifiable system defect with downstream engagement consequences. Alexa Economics projected that closing the latency gap across high-priority voice intents would generate $50M in incremental operational value over 15 months — a third-party validation that became the foundation of Alexa's Latency North Star.
NVA Model
Source: Alexa Response Latency Research — Devices Design Group, April 2021
Where cognitive error costs lives, not conversions
Translating trust frameworks to human-machine interfaces
Evolving established models into modern Human-AI trust frameworks