Skip to content
Enjoy some plunky browse music :)

Case Study · 2023–2026

Level Up

How a campus tour at Google turned into a custom-built LMS, a four-week intensive on career readiness, and my first time at the front of a classroom.

Next.jsPrismaGemini APICloud RunTailwind CSSNextAuth

At a glance

What
A production LMS built for my AI career-development course
My role
Built the platform, teach the course
Outcome
Live at levelupqu.com, powering the course
Stack
Next.js · Prisma · Gemini API · Cloud Run

The Spark

In the summer of 2023, Quinnipiac's "QU in LA" program brought a group of students to Los Angeles. I'd just finished my MS at Quinnipiac — most of it completed online, from my apartment in LA — and I volunteered to host them at Google's Playa Vista campus.

I put together two panels: one on early career alignment, one on leadership paths. Got Krista Phillip, a QU alum and product lead at Google, to join. Toured them through the Spruce Goose hangar, this old airplane manufacturing facility that Google turned into one of its most impressive offices.

I expected a nice afternoon. It turned out to be more than that. The students were sharp, full of questions I remembered asking myself not that long ago. One of them later called it "one of the best parts of the summer."

That day stuck with me. Not just because it went well, but because of a gap I kept noticing: these students were smart and motivated, but nobody had taught them the practical, strategic work of actually building a career.


The Partnership

From Tour Guide to Professor

Over the next year, I kept in touch with Andres Rosende, the program coordinator at Quinnipiac's School of Communications. We kept circling back to the same problem: students were graduating with strong academic credentials but limited practical tools for actually landing a job. The messy, human, strategic work of building a career wasn't being covered.

What if I didn't just show students around Google for an afternoon? What if I designed an entire course?

Andres and the school gave me the green light. I would design the curriculum, build the platform, and teach it myself as an adjunct professor during the Spring 2026 QU in LA cohort.


The Course

Four Weeks, Four Modules

I built the course around two books that changed how I think about careers: Meg Jay's The Defining Decade and Robert Cialdini's Pre-Suasion. Four weeks. Each module built on the last. Every part had readings, written reflections, and assignments that forced students to actually do the thing, not just learn about it.

Building Your Brand

Week 1

LinkedIn bio, identity capital audit, professional headshot, resume update. Everything posted to the class discussion board for peer review.

Building Your Network

Week 2

Weak ties, coffee chat openers, 3 real LinkedIn outreach messages, a 30-minute networking meeting with a professional, and a handwritten thank-you letter.

Building Your Impact

Week 3

Head-up vs. head-down work, pre-suasion, and a 12-minute efficiency presentation delivered at Google with live feedback from five engineers.

Building Your Start

Week 4

AI-driven interview practice using the STAR method, creative application strategies, and a comprehensive Personal Action Plan as their final deliverable.

The final deliverable was a comprehensive Personal Action Plan: top 10 target companies, springboard job listings with overqualification justifications, a professional bio, 5 STAR interview stories, a cover letter template, references, and an accountability reflection. Not an academic exercise. A living career document.


The Platform

Why I Built My Own LMS

I could have used Canvas or Blackboard. But those platforms are built for broad university administration, not for a four-student intensive with AI-driven interview practice and automated assignment analysis. I needed something specific, so I built it.

levelupqu.com is a production Next.js application I built from scratch. It runs on Cloud Run, stores data in PostgreSQL via Prisma, and integrates Gemini for real-time AI features. Three things it does that no off-the-shelf LMS could:

AI Interview Practice

Students paste a real job posting, and the system generates behavioral questions. They respond out loud using the Web Speech API, and Gemini evaluates their answers for STAR method structure. A comparable commercial tool runs about $25/month.

AI-Assisted Grading

When a student uploads a reflection PDF, Gemini analyzes it against the assignment rubric. I still read everything myself, but the AI gives me a content summary and suggested feedback points. Useful when you're grading at midnight between meetings.

LevelUpBot

A course-aware chatbot for logistics: "When is the residency?" or "What's the PAP due date?" I engineered the system prompt to refuse anything academic. It won't summarize readings or interpret course concepts.


The Classroom

First Day, First Cohort

February 4, 2026. Four students.

In my welcome email, I told them straight: "To be totally candid: this is my first time teaching, and I couldn't have asked for a better inaugural cohort."

We met Wednesday evenings at 6:30. Some weeks in person, some virtual — I was juggling my day job at Google and travel. I learned quickly why professors take forever to grade things. As I put it in one of my weekly emails: "Life gets you fast."

The first week, I assigned the preface and opening chapter of The Defining Decade — the one on "Identity Capital." By week two, they were reading about weak ties and sending real LinkedIn messages to professionals they'd never met. By week three, they were practicing pre-suasion techniques and building a pitch presentation. The pace was fast. That was the point.


The Residency

February 27 at Google

Google's Playa Vista campus. 8:30 AM to 5 PM. A full-day residency I'd been building toward since the first class session.

I recruited five Google colleagues to run three panel blocks. Early career insights. Leadership perspectives. And then the big one: a judging panel where the students would pitch their capstone project and get live feedback on their logic, ROI math, and presentation skills.

The students presented "Stop Sleeping Your Day Away" — a process improvement pitch for an app called Sheep Counter. They came in with real data: 56% of people hit snooze, their group loses about 2 hours every morning getting ready. They built a working prototype, scaled their ROI math to Google's 187,000 employees, and calculated over a million minutes of productivity gains.

Four weeks earlier, most of them hadn't heard of the STAR method. Now they were standing in a Google conference room, pitching to engineers, fielding questions about their data, and walking out with feedback they could actually use.


In Their Words

What They Took Away

Right after the residency, I asked each student to fill out a reflection survey. These are their words, unedited.

"No shade to anyone, but we took a class at Quinnipiac that was meant to give us similar tools, but this class was handled much more effectively."

— Post-residency survey

"It's crazy to think this course was only four weeks long, and yet, I feel like I've learned an entire semester's worth of important knowledge and skills."

— Post-residency survey

"It is probably the most I've felt like an 'adult' in my life and showed me things I didn't know I was capable of."

— On the Google residency

"I have never felt more prepared to begin building my career in the way I have always desired."

— Personal Action Plan


Under the Hood

Engineering the LMS

Two problems I ran into while building the platform, and how I solved them.

Timezone Integrity

Problem: Due dates stored as "11:59 PM" kept mismatching between the server (UTC) and students' local time in LA.

Solution: Standardized the database on UTC. Built a React bridge that converts datetime-local values to ISO on submission and re-localizes them for the student view.

Syllabus-to-Context Pipeline

Problem: The chatbot needed to stay current as I tweaked the course week-to-week.

Solution: A "Context Notebook" architecture where the bot's system prompt gets dynamically injected with the latest data from the Module, Assignment, and Material tables in PostgreSQL.


Tech Stack
Next.js 14
PostgreSQL
Gemini API
Cloud Run
Prisma ORM
Tailwind CSS
NextAuth.js
Web Speech

By the Numbers
4Students, 4 weeks
5Google panelists
1Custom LMS from scratch

Looking Back

What I Learned

In my farewell email to the cohort, I wrote: "You weren't just students to me; you were an inaugural cohort that taught me so much about how to lead, how to listen, and how to stay 'Head-Up' even when the schedule gets busy."

I posted about the experience on LinkedIn. One of the Google volunteers commented: "Rewarding experience to interact with brilliant peers and students."

This project sits at the intersection of everything I care about: building software that solves real problems, teaching people things that matter, and doing it all while holding down the day job. The LMS is production code. The curriculum is mine. The students are real people with real careers ahead of them.

The last thing I told them was from the book we'd spent four weeks reading: "It's up to you to define your decade."


Course Design, Full-Stack Development & Instruction · 2023–2026

Next ProjectUC California Climate StewardsA Design Thinking engagement for UC Agriculture & Natural Resources: a Community of Practice platform for 9,000+ stewardship alumni.