Dillon Chi

Dillon Chi

UX Researcher and AI Generalist

Services → About →

Mixed Methods UX Researcher · AI Generalist

Hi! I am Dillon.

I help teams understand their

Dillon Chi — UX Researcher & AI Generalist

Recovering Wikipedia addict, underdog cheerleader, and perpetual non-profit volunteer.

Open to roles & contracts

Hybrid & remote · NYC area

Based in Queens · Brooklyn, Manhattan & beyond

Let's work together
01

Selected work

Enrollment & Marketing Funnel Research

Enrollment & Marketing Funnel Research

Stride K12

edtech saas

Research supporting virtual public-school enrollment and in-product apps across a B2B2C EdTech platform.

  • Lead Funnel Conversion & Experimentation — Tracked and analyzed 12+ marketing funnels and hundreds of pages using Heap. By developing a bespoke set of metrics to prioritize rapid VWO experiments across layout, copy, and required-field indicators, we bypassed friction points to systematically improve user experience and discovery.
  • Mixed-Methods Research & Interactive Prototyping — Blending qualitative screeners and automated audits to scope a major redesign. We surveyed visitor intent across 100+ sites using Qualtrics, synthesized cross-network priority datasets in Tableau, and scraped mobile breakpoints with Playwright. To validate the solutions, we built interactive mobile prototypes in Replit because traditional prototyping would have been clunky and time-consuming. I also built bespoke tools to organize and map the Information Architecture (IA) of 70+ school sites in a single, centralized dashboard.
  • End-of-Funnel Conversion & LCP Tuning — Rescuing lost revenue in the K12 Enrollment Portal. By pairing a new approach of granular event micro-tagging in Heap (tracking actions rather than simple page-level analytics) with throttled mobile QA and Figma UX audits (validated against Salesforce drop-off data), we optimized document-generation LCP from 3–30s to under 2 seconds and unlocked a +9.9pp conversion increase.
Open Case-studies
Non-profit Web Presence & Strategy

Non-profit Web Presence & Strategy

Vermilion Sea Institute

conservation full-stack-automation

Technical stewardship and digital modernization for marine conservation.

Open Case-studies
Community Wellness Platform

Community Wellness Platform

Cedars-Sinai / ArtCenter

population-health social-good

Research to support the development of a fitness/research platform to conduct population science in Los Angeles in collaboration with Cedars-Sinai, Cancer Research Center.

Open Case-study
Auditing Hiring Algorithms for Fairness

Auditing Hiring Algorithms for Fairness

Speculative / ArtCenter

machine-learning systemic-bias

90% of Fortune 500 applicants are judged by a machine (an ATS). A toolkit to start testing those judgments and make hiring fairer.

  • 700 equally-skilled résumés — A controlled corpus for bias testing.
  • ATS judgment testing — See how the machine ranks identical skill.
Open Case-study
Real Estate Marketing Software Research

Real Estate Marketing Software Research

Listreports / Highway.ai

real-estate-software research-operations

UX Research at a scrappy startup, I wore many hats.

  • Reactivation Flow Optimization — Streamlining B2B subscriber reactivation. I mapped operational handoffs between CS and Engineering, used Heap to analyze reactivation drop-offs and isolate manual bottlenecks in Salesforce account provisioning, and built an automated self-serve billing step to convert 300 ghost subscribers back into paying users.
  • Predictive Engagement Modeling — Defining what active usage means at scale. I applied the Reforge methodology alongside probability models and regression analysis to classify 300k+ users into distinct lifecycle cohorts, providing product teams with clear quantitative thresholds to predict subscriber retention.
  • Cross-Product Event Instrumentation — Designing B2B product analytics from the ground up. Post-acquisition, I built the full tracking infrastructure for 20+ legacy products using a systematic naming hierarchy, deployed Heap and Mixpanel dashboard suites, and established a data-driven framework to sunset or invest in products based on active user share.
Open Case-studies
Niantic-Commissioned LED Sculpture

Niantic-Commissioned LED Sculpture

Niantic Labs Commission

interactive-sculpture led team-leadership

One of thirteen chosen proposals out of hundreds — an 8-petal LED sculpture commissioned for an event sponsored by Niantic Labs (Pokémon Go creator).

Open Case-study
Immersive VR Sculpture Installation

Immersive VR Sculpture Installation

ArtCenter College of Design

vr sculpture 3d-modeling

A final project exploring intersecting volumes and immersive spatial interaction — from cardboard prototype to a Unity VR installation participants could step inside.

Open Case-study
Consulting

Consulting

Available for engagements

insight-gathering analytics workflow

Qualitative and quantitative insight, analytics, and workflow optimization — built around your business goals and your stakeholders.

Open Case-study

Online AI Exhibition

Is AI inherently biased
if it is trained on us?

Labels
Cropped photo of a hand holding a thermometer gun, classified by an AI model Same photo cropped differently, now classified as a monocular by the AI model
Hand 77%
Gun 61%

Algorithmic bias

"It saves money to let robots do human things"

But can we afford to?

The same image — a person holding a thermometer — gets read two different ways by the same vision model. First it sees a gun. Crop the frame slightly, and it sees a monocular instead. Neither is right. Both carry consequences when systems like these get deployed at scale, on real people, with real stakes.

Do humans do robot things?

Every CAPTCHA you solve teaches an AI system to recognize.

Train the machine

Are they robots?

Self-driving car dashboard view
Ride-share app interface
Smart home delivery
Amazon warehouse automation
AI-powered interview screening
Automated interview process

"Gets me from point A to B and I don't even have to talk"

Or are they humans?