Paramount+ internship · Summer 2026

I built Verdict, a game that helps a room pick a movie.

A side project I took on myself, as both the PM and the engineer. The TV hosts, everyone votes from their own phone, and nobody installs anything.

Working prototypeDemoed to Paramount leadership

Visa · Visado Category / Categoría  Product Issued at / Expedido en  Paramount+ · New York
Muted loop of the demo.Watch the demo
  • 6 weeks,idea to demo
  • 6 sessionswith a VP of Product
  • Playtested by8 interns
  • Demoed toan SVP

How it works

4 steps · Pasos
  1. TV lobby: Scan to Join with a QR code, room code S5G3C2, and eight player slots with Player 1 joined.

    Start. The host taps "Start a game" on the TV, and a QR code appears.

  2. Phone browser: What's the mood? Pick up to three, with options such as Cozy & Comforting, Edge-of-Your-Seat, Mind-Bending and Big & Epic, each with an example title.

    Join. Everyone scans it, picks a nickname, and chooses up to three moods.

  3. TV: The Verdict Is Coming. Everyone's swiping, top votes make the podium, with a progress bar under each player.Phone: a movie card being swiped right with a green Like label, and the next card underneath.

    Swipe. Everyone swipes the same 10 movies on their phone.

  4. TV podium: Pick Tonight's Winner, ranked by the room's votes. Babylon is first with 6 votes.Phone: Ranking locked. Eyes on the TV. The podium is loading.

    Pick. The TV shows the top three. Tap the winner to start watching.

    If it's a tie, everyone ranks their top three in a final round.

Verdict full demo, 1:19, with captions.Open on Vimeo

My PM work

Seven things I did as the PM, in order.

01 Endorsement · EndosoPM skill: Problem framing

I found the problem

My friends and I once scrolled for a movie so long that it ruined the night. This happens a lot. U.S. viewers spend about 12 minutes deciding what to watch (Nielsen, 2025). The catalog wasn't the problem. One person holds the remote, and nobody wants to be the one who picks.

12 minDeciding what to watch
02 Endorsement · EndosoPM skill: User focus

I designed for the guest

The host has the Apple TV and the subscription. Everyone else just has a phone, and it might be an Android. So guests join from their browser, with no app and no account.

Host

Has the Apple TV and the subscription.

Guest

Joins from their browser, with no app and no account.

03 Endorsement · EndosoPM skill: Iteration

I shaped it with a VP of Product

I met six times with Paramount's VP of Product Management, Streaming Innovation. We talked through the overall design, where the game should live, and how people would actually use it. Those talks moved Verdict from Tinder-style swiping between phones to a Jackbox-style game on the TV.

Before: a phone swiping a movie card.Before: a second phone swiping, to match with the first.
Before
After: one TV showing every player's swiping progress while the whole room plays.
After
04 Endorsement · EndosoPM skill: User research

I tested it with users

I ran a playtest with about eight interns and took notes by hand. Their biggest ask was picking by mood instead of genre, so I rebuilt the picker around moods. I also demoed it to senior leaders from several teams and a family at home.

"I can totally see this changing the way I browse content."
"I wish I had this for my last movie night."
Handwritten notebook page titled Verdict Comments, listing playtest feedback such as narrow down genres, examples of genres, make it more obvious that you swipe, undo on swipe, flip to see description, by vibe with emojis, and rescramble if all say no.
05 Endorsement · EndosoPM skill: Prioritization

I cut the scope

I had six weeks, so I kept only what gets a room from "what should we watch?" to a pick. I said no to accounts, an iPhone-only app, and AI recommendations. Each one would have turned Verdict back into a regular streaming app.

Built

  • Mood picks
  • A 10-card deck
  • Live voting
  • A tiebreak round
  • The podium
  • Tap to play
  • Flip for description

Next

  • A veto card
  • Undo a swipe
  • Reshuffle when everyone passes

Said no

  • Accounts
  • An iPhone-only app
  • AI picks
06 Endorsement · EndosoPM skill: Metrics and risk

I defined success

Verdict didn't ship, so these are the numbers I'd launch with.

I also planned for what could break it. Spotty Wi-Fi falls back to the local network, ties go to a final round, and the game only stores a nickname, which is deleted when the game ends.

North star

Games where the winner starts playing within 5 minutes

Watch

How many guests join, and time to a pick (goal: under 90 seconds)

Guardrail

People still use My List and watch as much as before

07 Endorsement · EndosoPM skill: Communication

I pitched it to leadership

In my last week, I demoed Verdict to Paramount's SVP of Innovation and Emerging Platforms and the VP of Product Management, Streaming Innovation. They said it filled a real need and saw a market for it. They said I shouldn't be afraid if this ends up showing up on my screen one day. It was my last week, so the project ended there.

The podium reveal shown in the leadership demo: Babylon wins tonight's pick.

What I'd do differently

I built this a bit siloed. Next time, I'd do user research before I start building, so the idea isn't just living in my own head.

How I built it

SwiftUI on the Apple TV, a plain web page for phones, and Firebase to pass votes between them. I used Claude Code to build fast, but the product calls were mine. When it pushed an Apple-only design, I kept it open to Android, because most living rooms aren't all iPhones.

Technical details
Bearer / TitularPassport / Pasaporte
Shelby Lovejoy, smiling, in a black blazer, leaning against a window outside a café.

Shelby Lovejoy

Studying
Computer Science, Georgia Tech · May 2028
Threads
Information Internetworks, Human-Computer Interaction
Now
Product Fellow, Product@GT x UPS (LATAM)
Previously
Paramount+ Apple team
Toolkit
Figma, Jira, SQL, Swift, Python, React

Other work

Feature Flag Health ReportA tool that found 90 dead or stale flags out of 385. Two other teams asked to use it.
Crash-triage automationA six-step workflow that replaced a manual pass every sprint, with a human approval step at the end.