Dubai-raised Β· Michigan-made Β· AI Engineering
ZANE
LUIS
Building AI systems that actually ship.
Exploreβ 1,800 particles, one draw call
Fibonacci-sphere distribution instead of random placement β uniform density with zero clustering. Rendered as a single THREE.Points buffer (one draw call), hue cycled in the render loop instead of re-uploading vertex colors. Deferred via dynamic import so Three.js never blocks first paint.
const phi = Math.acos(1 - 2*(i+0.5)/n);
const theta = Math.PI*(1+Math.sqrt(5))*i;Last commit
β
14-day commits
0
Chapter 01 β Origin Story
The Journey
Dubai to Michigan. Two cities, one trajectory.
The Beginning
Dubai
UAE
GEMS Modern Academy
2012 β 2025
βWhere curiosity met code. Top of class, captained the basketball team, taught myself to build things, and discovered a love for turning ideas into software.β
The Mission
Ann Arbor
Michigan
University of Michigan
2025 β 2029
βThe AI chapter. One of the best CS programs in the country, two production AI systems shipped. The mission: become an AI engineer who builds systems that scale, matter, and ship to millions.β
Chapter 02 β What I Build
Built to ship.
Production-grade systems, not demos. Every architectural decision was deliberate.
A full-stack dating platform with a multimodal AI compatibility engine that scores matches on four signals using Claude Vision, Jaccard similarity, and personality alignment.
Multimodal AI pipeline: 40-label controlled vocabulary via structured prompting β zero label drift
Four-signal compatibility score: Jaccard (40%) + personality (20%) + facial match (30%) + mutual bonus (10%) β O(n) per request
Production backend: 6 domain routers, JWT auth, email verify, SQL block filtering, 200 req/min rate limiting
Real-time WebSocket chat, Framer Motion swipe UI with rAF pause on visibility loss, PWA installable
Chapter 03 β The Craft
The craft.
Tools I use to ship. No fluff.
β The whole site's colors are 3 CSS variables
Every gradient, cursor, scrollbar, and border reads --a1/--a2/--a3. Registered via CSS @property so colors interpolate natively β an IntersectionObserver swaps the palette per chapter and the browser tweens everything. Zero re-renders, zero JS animation loops for theming.
@property --a1 { syntax: "<color>"; }
:root { transition: --a1 0.9s ease, ... }
// IntersectionObserver sets vars per chapterChapter 04 β The Reps
The reps.
Data Analytics Intern
Samsonite
- Analyzed sales performance data to evaluate growth opportunities for the American Tourister Airconic product line
- Assessed target markets, pricing ranges, and consumer segments to inform product development strategy
- Built visual dashboards in Google Looker Studio to communicate findings to stakeholders
Product Development Intern
Cult.fit
- Designed a fitness mobile app MVP aimed at making health and wellness more accessible
- Conducted market research to identify user pain points and define a clear problem statement
- Translated insights into a user-centric MVP design
Remote Java Coding Instructor
Kyangala Girls High School, Kenya
- Taught Java programming remotely to underprivileged students as part of a Duke of Edinburgh Award volunteering project
- Designed and delivered lessons covering programming fundamentals, logic, and problem-solving
- Several students expressed interest in pursuing STEM β bridged geographic and cultural gaps through sustained remote mentorship
Head of Web Development
Debuggers Club
- Led development of a centralized database system to digitize the school House Points System
- Built and maintained full-stack web applications (HTML, CSS, React + PHP, MySQL, Apache)
- Designed a device tracking system for school-owned hardware
Member β Machine Learning Club
University Machine Learning Community
- Built a Python text sentiment classifier using Keras at the Cursor & MCP Showcase
- Participated in workshops on ML fundamentals, clean code structure, and rapid prototyping with AI-assisted tools
Chapter 05 β Beyond Code
Play as Zane.
Three playable moments from my life before code: captaining the varsity basketball final, defending a U18 swimming title, and learning FΓΌr Elise on the piano. Same instincts that make an engineer β timing, pattern recognition, iteration.
Chapter 06 β The Algorithm, On You
Match with Zane.
This is the actual four-signal compatibility engine from Lovemaxxing β Jaccard similarity, weighted signals, normalized scoring β running on you, right now, in your browser.
score = 0.4Β·J(A,B) + 0.2Β·P + 0.3Β·F + 0.1Β·M
β Same math as production Lovemaxxing
This isn't a mock β the weights (40/20/30/10) and Jaccard set-similarity are the same signal design as the deployed dating platform, which ranks candidates O(n) per request. The only difference: there, signal 3 compares AI-detected facial features against stated preferences via a 40-label controlled vocabulary.
function jaccard(a, b) {
const inter = [...A].filter(x => B.has(x)).length;
return inter / new Set([...a, ...b]).size;
}Ready to get scored?
5 questions. Real algorithm. No feelings spared.
Chapter 07 β What's Next
The Zane Exchange
PAPER TRADINGInspired by my Kalshi NBA bot β bet on my career instead. Prices move with your trades.
Zane ships his next AI project
by Sep 2026
Lands Summer 2027 AI Engineering internship
by Mar 2027
NBA prediction bot goes live on Kalshi
by Dec 2026
* no real money. but I am genuinely long on all three.
Let's build something.
I'm actively looking for Summer 2027 AI Engineering internships. If you're working on something that matters, I want to hear about it.
Β© 2026 Zane Luis Β· Ann Arbor, MI
zaneluis.com