UX Researcher & Human Factors Data Engineer

Sebastian
La Rosa

UX research and human factors for AR/XR — from user studies to CAD & scan analysis

Sebastian La Rosa — front
Sebastian La Rosa — lateral

About

Research at
the edge of perception

I am a UX Researcher and Human Factors Data Engineer at Magic Leap, conducting human-centered research across rapid prototyping and hardware development environments for AR/XR head-mounted display systems. I work across AR/XR platforms and Magic Leap reference prototypes, and collaborate with a partner company on cross-organizational research.

My work bridges behavioral UX research and physical human factors — contributing across study design, user interviews, data analysis, and synthesis as part of the research team, while owning scan analysis from capture through reporting. Across 2,000+ human subjects, 600+ scan alignments, and multiple device generations, I've developed new methods adopted across the research team and co-authored published research on nose shape and HMD design, with a second paper in progress. BBA from Florida Atlantic University. Certified in GCP for Human Subject Research.

3.5+Years in XR Research
2,000+Human Subjects Studied
$2.2M+Study Cost Savings
600+Scan Alignments

Experience

Magic Leap

UX Researcher & Human Factors Data Engineer · Mar 2023 – Present · Plantation, FL

  • UX, HCI, and Human Factors research across rapid prototyping and hardware development for AR/XR HMD systems and platforms, including Magic Leap reference prototypes.
  • Collaborate with a partner company on cross-organizational research across a variety of projects.
  • Led scan analysis across multiple external studies including Proof of Concept efforts — evaluating comfort, fit, slippage, and visual registration across multiple glasses prototypes. 600+ total alignments; developed new visual registration, pupil determination, CER, pitch delta, and clearance-map procedures now adopted by the research team.
  • Helped design and run remote, unmoderated diary studies capturing longitudinal AR experience — including perceived cognitive comfort, mental load, and day-to-day wearability — through activity design, survey instrumentation, data cleaning, synthesis, and stakeholder reporting.
  • Research contributions across multiple cross-functional studies have informed product, design, and engineering decisions — including a team proof-of-concept that exceeded KPIs and supported an estimated $2.2M+ in projected cost savings.
Show full experience
  • Experience presenting research findings to cross-functional stakeholders including product, engineering, design, and leadership.
  • Contributed to comfort and fit studies — fitting participants into multiple prototypes and device iterations (varying nose pieces and sizes), conducting 2–4 hour all-day wear experiments, and evaluating slippage, virtual FOV, and visual registration in session.
  • Supported benchmarking and focus-group studies comparing AR/XR competitor devices against internal hardware — hands-on evaluation, proctoring, social-acceptance assessment, and feedback synthesis.
  • Applied AI-assisted research methods (Gemini, NotebookLM) for competitive analysis, lit review, and research proposal drafting; built an API-based data pipeline for user sentiment at scale.
  • Built and delivered the research team's PolyWorks Scan Analysis training program — structured materials, guided working sessions, and hands-on coaching that significantly improved team proficiency and consistency. Acts as the team's technical expert for scan analysis: QA, mentorship, and resolving complex issues with reliability. Presented the full methodology in a formal Shareout to UX Research and Product teams.
  • Developed personas, journey maps, and executive research summaries communicated to product, engineering, and leadership stakeholders.
  • Co-authored a peer-reviewed paper on nose shape categorization and HMD design; credited on a second paper currently in progress.

Mars Research

Market Research Intern · Jan 2023 – Mar 2023 · Fort Lauderdale, FL

  • Early-career research internship — market research, competitor analysis, A/B testing, focus groups, and survey-based studies supporting client strategy.
  • Analyzed qualitative and quantitative data and built executive summaries and visualizations communicating findings to clients.

Florida Atlantic University

BBA Business Administration, Management

Published Research

Peer-Reviewed Publication · Co-Author

Nose Shape Categorization & Its Impact on HMD Research

Evaluates how Martin & Saller's nasal index correlates with comprehensive nose shape variables using Principal Component Analysis (PCA) and nonparametric bivariate correlation analysis in the context of HMD design. Published in an ergonomics & HMD design proceedings, 2024. Co-authors: Tom Schnieders & Karen Bredenkamp.

A second paper on nose shape categorization is currently in progress — credit pending publication.

Read Paper →

Portfolio · Part 1

UX Research
Portfolio

A selection of UX research work across AR and XR. Due to NDA and confidentiality requirements, the case studies below represent a redacted cross-section of work that also spans AR navigation, spatial content placement, OS and home menu design, notification systems, display and dimming research, multiple device application experiences, voice and gesture interaction, and system communications across multiple device generations.

Platform Research

AR/XR Platforms · Reference Prototypes

Platform & Device Research

UX research across AR/XR head-mounted displays and Magic Leap reference prototypes — spanning interaction design, input modalities, OOBE and first-use flows, onboarding, and platform-level decisions across hardware and software layers.

Benchmarking

Device Benchmarking

Multi-round benchmarking studies comparing AR/XR interaction modalities and competitor hardware — spanning hand tracking vs. controller performance (SUS, perceived comfort, accuracy scales), hand anthropometry collection, arm and wrist fatigue assessment, and competitive device evaluation across industry-leading XR platforms.

AI-Assisted Research

AI Research Methods

Applied Gemini and NotebookLM for lit review, competitive analysis, and proposal drafting. Built an API-based data pipeline to extract and analyze user sentiment from public platforms at scale — delivered as written proposals to stakeholders.

Product OS & App Research

On-Device Experiences

Evaluative and generative research across on-device AR experiences — home menu and OS navigation, notifications and system communications, display and dimming, spatial content placement, app usability (capture, scanning, onboarding, developer apps), and AR map navigation. Work spans behavioral and attitudinal methods across both qualitative and quantitative approaches, informing product road maps and key design decisions.

Case Studies

Benchmarking · Mixed-Method

Hand Tracking — OS Navigation

Can people navigate an AR operating system with their hands alone — well enough to ship it on by default? A 28-participant mixed-method benchmark across the full task suite, and the rating-scale instrument behind it.

Usability TestingBenchmarkingSUSHand TrackingAR / XR
Read case study

The Challenge

Direct-interaction hand tracking was a candidate to enable by default — but it was unknown whether first-time users could complete core tasks (device setup, scrolling, text entry, sliders, browsing) without frustration or fatigue.

My Contribution

I contributed across protocol and rating-scale design, moderation, behavioral coding, and synthesis as part of the research team.

Methods

28 external, native users across a within-subjects task suite. Mixed-method: SUS, structured observation, and per-task rating scales for discoverability, ease of use, perceived accuracy, and perceived comfort.

Method

Moderated, in-person · Within-subjects

Participants

28 external, native users

Instruments

SUS · Rating scales · Observation

Study Instrument — Per-Task Rating Scales

Discoverability

12345
Very difficult to discoverVery easy to discover

Ease of Use

12345
Very difficult to useVery easy to use

Perceived Accuracy

12345
Very inaccurateVery accurate

Perceived Comfort

12345
Very uncomfortableVery comfortable
76.0SUS — hand tracking
43.2 / 87.8Benchmarks — prior HT / controller
63%Participant pass vs. 90% target

Outcome

The benchmark drove a conditional-ship recommendation and a prioritized redesign brief — depth-feedback cursor, virtual keyboard, and gesture instruction — delivered to product and engineering.

Moderated Usability · Iterative

AR Onboarding & First-Use

How do first-time users learn to operate an AR device across hand, controller, and voice input? Iterative usability research that directly shaped Fundamentals — a Magic Leap onboarding app launched to teach the basics of AR.

Usability TestingOnboardingIterative DesignAR / XR
Read case study

The Challenge

Introducing someone to AR means teaching an entirely new interaction paradigm, with no established mental model to fall back on — across users with very different expectations, learning styles, and tolerance for frustration.

My Contribution

Contributed across study rounds — moderation, behavioral coding, and synthesis — as part of the research team.

Methods

Multiple rounds of moderated, think-aloud usability testing with first-time AR users, progressing from controlled lab sessions to real-world deployment environments — private offices, open factory floors, conference rooms, and homes — to see how the experience held up where users would actually encounter it.

From the Study — Direct-Interaction (Poke) Onboarding

My hand interacting with the Fundamentals onboarding app — poking virtual objects to learn direct interaction.

Outcome

Findings shaped each iteration of the app’s content, flow, and interaction design, including tutorial priorities and in-context guidance — the research behind the shipped onboarding experience.

Comparative Evaluation

Clipping Plane — Content Rendering

When virtual content meets the edge of the display, how should it disappear? A within-subjects study of three rendering conditions across real AR tasks, delivered as end-user and developer guidance.

Comparative EvaluationWithin-SubjectsPerceptionAR / XR
Read case study

The Challenge

The point where rendered content clips at the display boundary affects perceived solidity, depth, and visual comfort — but the “right” behavior wasn’t defined.

My Contribution

Helped design the study and contributed across moderation, persona coding, and synthesis.

Methods

37 external participants evaluated three conditions — baseline (hard clip), conservative (near-fade), and liberal (no clip) — across spatial object search, direct UI interaction, and close-range reading. Preference ratings, persona coding to segment participants, and a structured qualitative debrief.

Outcome

Findings revealed nuanced preferences around content solidity, perceived movement, and perceived visual comfort, and were formatted as end-user and developer guidance published to the AR developer portal.

Remote Diary Study · Longitudinal

Longitudinal Wearable Experience

What does it feel like to live with an AR device over time? I run remote, longitudinal diary studies — bookended by 1:1 interviews — to understand how people actually live with a device, day to day.

Diary StudyUnmoderatedUser InterviewsLongitudinal
Read case study

The Challenge

Some questions only answer themselves over time — how comfort, attention, and adoption evolve once a device becomes part of someone’s daily routine. A diary study is the right tool to capture that lived experience, alongside the moderated lab research it complements.

My Contribution

I help design and run these studies — activity design, survey instrumentation, data cleaning, longitudinal synthesis, and stakeholder reporting.

Methods

An unmoderated diary phase — participants self-reporting through structured prompts and surveys in their own environments — paired with 1:1 user interviews at the start and end, tracking perceived cognitive comfort, mental load, social acceptability, and perceived physical comfort over time.

Outcome

The longitudinal, in-context format gives a richer view of how people actually live with a device — how adoption and comfort shift with real-world use over time — complementing moderated lab research and feeding product and design decisions.

Portfolio · Part 2

Human Factors
Research

Our scan analysis process begins with in-person collection in our lab. The team runs anthropometry on each participant — landmarking key facial points by palpation, then capturing baseline and fit scans. After the session, I digitize those landmarks in PolyWorks, determine coordinate points, and measure the deltas between imported and aligned positions. Physical collection is imperative — scan analysis is what allows us to investigate findings further, giving the team extended time with a participant's anatomy long after they've left the building.

This generates findings that engineering and industrial design teams act on directly — verifying comfort and slippage observations from the live session, evaluating visual registration more accurately than in-person proctoring allows, and producing measurements that feed mechanical engineering decisions like clamp force.

This work directly supports decisions around physical comfort, slippage mitigation, and visual registration — the critical factors that determine whether an AR device can be worn all day by a diverse global population.

Presenting scan analysis to cross-functional stakeholders

Scan alignment & analysis — presenting to cross-functional stakeholders

Additional Work — Device Fit & Comfort Studies

Beyond scan analysis, I contribute to large-scale physical fit and comfort studies evaluating AR/XR HMDs across diverse participant populations. Our team draws on a 2,000+ person anthropometry database — 19 traditional 1D measurements, 18 facial landmarks, and sub-millimeter-accurate 3D head scans across Asian, Caucasian, and African-American demographics — the statistical foundation for fit validation at population scale.

In-person sessions combine anatomical landmarking by palpation (glabella, tragion, nasion, infraorbitale, zygion), traditional anthropometric measurements (head breadth, face breadth, nose widths, IPD, bitragion breadth), and 3D scan capture. Comfort is rated on 5-point Likert scales every 30 minutes across 2–4 hour sessions, slippage via a proctor-observed protocol, with FOV and eye-tracking registration captured in parallel — findings feeding industrial design, mechanical engineering, and product decisions.

Step 1 — CAD Input

Every analysis starts from two physical inputs, both captured in our lab. The device's CAD model is 3D-scanned on its own, and the participant is scanned wearing the device during a fitting session. The CAD geometry represents design intent; the participant scan represents physical reality on a real human head. (The frame shown here is a non-proprietary example used purely to demonstrate methodology.)

CAD reference views

CAD view 1 CAD view 2 CAD view 3 CAD view 4 CAD view 5
Step 2 — CAD-to-Scan Alignment

The device CAD is aligned to the participant's fit scan — taken with the device worn — and compared against the participant's baseline head scan (captured separately wearing a wig cap). This reveals exactly how the device seats on that individual's facial geometry, and whether it falls within the intended design envelope. Below: one participant with a successful fit, one with a population mismatch.

Good Fit
Good fit scan Good fit baseline

Fit scan + baseline head scan — natural seating, frame within envelope

Poor Fit
Bad fit scan Bad fit baseline

Fit scan + baseline head scan — frame elevated, sits outside contact zones

Step 3 — Contact Mapping

Contact maps quantify skin-to-frame contact at every surface point, color-coded by proximity. Blue = no contact (frame held away from the head). Green = normal, light contact. Red = high-pressure contact, where the frame presses into the head. Together with the in-person session data, these maps let me investigate exactly where and why a fit succeeds or fails.

Contact
No ContactNormalHigh Pressure
Good Fit — Contact Maps

Good fit views

Good contact 1 Good contact 2 Good contact 3 Good contact 4
Poor Fit — Contact Maps

Poor fit views

Bad contact 1 Bad contact 2 Bad contact 3 Bad contact 4 Bad contact 5
Research Outputs
01
Comfort & Discomfort

Scan analysis verifies and deepens the comfort findings gathered during the in-person fitting session — confirming where contact concentrations align with the discomfort participants reported, with precise anatomical location data for the engineering team.

02
Slippage

By showing exactly how the temple arms and nose pieces contact the head — the interfaces that actually hold the device in place — contact analysis helps explain the slippage behavior observed in person, grounding session findings in measurable geometry rather than proctor estimation alone.

03
Visual Registration

Using the eye-box CAD linked to the frame, I calculate center of eye rotation (CER), pupil position, and eye-box placement from digitized anthropometric landmarks — the geometry that governs where virtual content lands in the visual field, how sharply it resolves, and the binocular alignment that drives visual comfort and reduces eye strain. Can be more accurate than in-session proctor evaluation.

04
Cross-Section & Angle Analysis

Cross-sectional cuts at key anatomical landmarks reveal internal clearance geometry not visible from surface maps. Temple arm opening angles and pitch delta are measured across participants and compared against the population database — data the mechanical engineering team uses directly for clamp force and structural fit decisions.

Methods & Toolkit

Research
Capabilities

UX Research

  • Moderated Usability Testing
  • 1:1 User Interviews & IDIs
  • Contextual Inquiry & Field Studies
  • Diary Studies
  • OOBE & First-Use Research
  • Concept Testing & Comparative Evaluation
  • Gesture & Voice Elicitation
  • Survey Design · SUS · Custom Rating Scales
  • Persona & Journey Mapping
  • Participant Recruitment & Proctoring
  • Competitive Benchmarking
  • AI-Assisted Research (Gemini, NotebookLM)
  • Logs Analysis

Human Factors

  • 3D Fit Scan & CAD Alignment
  • Contact Pressure Mapping
  • Cross-Sectional Analysis
  • Anthropometric Data Collection
  • HMD Donning & Fit Validation
  • Physical Comfort & Fatigue Assessment
  • PolyWorks Inspector Analysis
  • JMP Boundary Case Selection
  • 3DMD Wrap / Vultus Processing
  • Principal Component Analysis (PCA) · Statistical Analysis
  • Large-scale Human Subject Research
  • Visual Registration Methods
  • New Method Development

Domains & Tools

  • Augmented Reality (AR)
  • AR/XR Platforms
  • Magic Leap Reference Prototypes
  • Head-Mounted Displays
  • Smart Glasses & Wearables
  • Dev App Research
  • Rapid Prototyping Labs
  • Cross-org Matrixed Environments
  • Figma · Jira

Contact

Human behavior
is the data.

Open to conversations about UX Research, Human Factors, and Spatial Computing. Based in South Florida.

Let’s talk →