Devices Design Group
UX Researcher
Calibrating Alexa Latency Targets
to Human Perception
Incremental OPS prevented by fixing latency thresholds.
Engineering targets for Alexa were historically arbitrary. I led a multi-year research program to replace legacy technical benchmarks with human-centered thresholds, directly linking millisecond delays to long-term revenue loss.
Experimental Design: Human Factors
Used a custom "Wizard of Oz" latency engine to manipulate response times with millisecond precision across 2,160 controlled interactions with Alexa. This rigorous mapping identified the "High Satisfaction Threshold" as all responses faster than 1000ms, where user satisfaction is highest and perception of "slowness" is minimized.
Strategic Integration: Economics
Partnered with Alexa Economics to integrate these perceptual thresholds into the Negative Value Action (NVA) model. We redefined "system defects" not just as technical failures, but as any interaction slower than the 1000ms satisfaction cliff—the point where the probability of customer churn and engagement loss sharply increases.
Outcome: $50M+ Projection
The calibrated model forecasted that achieving these human-centered thresholds for high-utility intents would save 50.3B dialogs with Alexa. Realizing these interaction gains was projected to generate $50M+ in incremental revenue over the subsequent 15 months, providing a definitive financial North Star for Alexa’s global engineering roadmap.
NVA Model
Fig 1. The "Satisfaction Cliff": User satisfaction drops sharply after 1000ms
Fig 2. Measuring Cognitive Load via fNIRS
Biometric
UI Benchmarking
Operationalizing "Clutter"
The FireTV Home UI design was historically based on aesthetic preferences. I shifted the conversation to cognitive cost. Using fNIRS (functional neuroimaging), we proved that FireTV's UI density triggered significantly higher activation in the Left DLPFC (working memory) compared to competitors (Apple TV & Netflix).
The Strategic Pivot
We optimized the Home UI to reduce cognitive load for 75M+ customers, resulting in increased engagement metrics and a simpler mental model for content discovery.
Multimodal AI
+ Hardware
Led end-to-end Human Factors strategy for the Echo Show portfolio. My research informed 50+ iterative design updates that improved the core user experience for 75M+ customers.
Fig 3. Echo Show 10 with Motion Tracking
Staff Product Researcher
Product Strategy Lead
Staff Product Researcher
Product Research Leadership
Lift in subscription conversion via checkout innovations.
Directing the research roadmap for monetization and Human-AI interaction, reporting directly to the VP of Product.
STRATEGIC FRAMEWORK: SLING EXPERIENCE INDEX (SXI)
Architected a proprietary quantitative framework (SXI) that merged behavioral telemetry with user-perceived friction, transitioning leadership to a proactive prioritization model now used for all VP+ product reviews.
STRATEGIC PIVOT: ROADMAP REALIGNMENT
Leveraging SXI longitudinal data, I orchestrated a 2025 roadmap realignment. I secured executive buy-in to halt low-value feature releases and instead prioritize a foundational overhaul of the checkout architecture to address critical friction hotspots.
MEASURABLE IMPACT
The re-architected checkout flow directly resulted in a 17% lift in subscription conversion during Q4 2025, validating the financial impact of the SXI framework in resolving systemic friction.
HUMAN-AI INTERACTION: ARCHITECTING TRUST
Led foundational research to define the multi-turn interaction model for a conversational TV assistant. Established performance benchmarks for latency and ambiguity resolution, and developed a "Trust & System Status" framework to align with customer mental models during complex, multi-turn tasks.
2025 Product & AI Workshop
Driver Growth & Global Strategy
UX Researcher
UX Researcher
Driver Retention Strategy
in Brazil
Increase in rental driver retention via app redesign.
Conducted foundational research in Brazil to inform global product strategy. Insights directly steered the driver app redesign, successfully reducing cognitive load and improving retention.
Foundational Research in Brazil
Conducted field research with rental drivers in Brazil to identify unique market constraints. Discovered that high cognitive load from the existing driver app was a primary churn factor for new rental drivers.
Business Impact
The implemented driver app redesign successfully reduced cognitive load metrics and resulted in a 5% lift in driver retention for the rental segment in the LATAM region.
Research Led to Driver App Redesign
Human Factors Research
Research Lead
Human Factors Researcher
Reducing Astronaut
Cognitive Load
Reduction in critical Time-on-Task for medical procedures.
Led Human Factors validation for next-gen medical workstations on the Lunar Gateway, minimizing critical operator errors in high-stress zero-G environments.
Validation: Lunar Gateway
Executed human factors validation studies for the Lunar Gateway medical workstation. This rigorous testing environment simulated zero-gravity constraints to identify physical and cognitive ergonomic failure points.
Operational Impact
The redesigned interface and physical layout resulted in a significant 30% reduction in time-on-task for complex medical procedures, ensuring astronaut safety and mission efficiency.
VR Simulation: Medical Workstation
Automotive UX & HMI
UX Researcher
UX Researcher
Trust in Semi-Autonomous
Driving
Increase in user trust during automated handoffs.
"How does the car communicate intent?" Conducted foundational research on passenger experiences in L2 vehicles to guide critical improvements to the HMI that enhanced trust.
HMI Design Standards
Defined Alert Modality standards for autonomous handovers. These findings directly influenced internal design guidelines, ensuring clear communication of system status during critical control transitions.
Psychoacoustic Modeling
Operationalized the subjective experience of "quiet" by correlating acoustic sensor data with human perception. This model guided engineering adjustments that reduced perceived cabin noise by 15%.
Research Impact
The implemented HMI improvements resulted in a verifiable 24% increase in user trust scores during automated lane changes and handoffs.
Semi-Autonomous Driving Simulator