Portfolio 2026USC Iovine and Young Academy

Building tech startups from the ground up.

Works as FounderAI prototyperProduct managerVisual designerFounder Scroll

My focus is entrepreneurship and creating tech startups from the ground up.

School
University of Southern California
Program
Iovine and Young Academy
Based in
Los Angeles
Selected work06
01TRIAGEAn AI waste sorter for hospital operating rooms. A prototype for a potential system.Co-founder
Triage, concept demo
02FRUIT.A venture studio that turns idle university research into student-built startups.Co-founder
fruit., concept demo
fruit. / IP library
  • Sample IP 01Biotech, patent
  • Sample IP 02Robotics, prototype
  • Sample IP 03Climate, concept
Team
Stage
IdeaValidationMarket
Concept validated
03CONTRIQAn AI agent that negotiates software renewals on the buyer's side.Founder
Contriq, concept demo
Contriq
Renewal notice

Your subscription renews next month.

Pricing will increase at renewal.

Reading termsBenchmarking, 20 vendors
Counter

We'd like to renew at current pricing.

In return, we can commit to a longer term.

04TAUNTA consumer AI app that holds you to your goals with escalating check-ins.Founder
Taunt, concept demo
9:15Taunt
TauntEssay due at midnight. You have written 0 words.
TauntTwo hours later. Still 0. Bold strategy.
TauntExcuse court is in session. State your excuse.
My laptop died.
VerdictOverruled. The library has outlets.
05WRAITHA wearable that masks the wearer's heat signature from thermal cameras.Co-founder
WRAITH, concept demo
Suit offSuit onThermal camera view
06TRACKTIONResearch tracing how songs move from TikTok to Spotify to the Hot 100.Co-lead
Tracktion, sample data lag TikTok videosSpotify streamsHot 100
01 / 06Featured projectCo-founder

A prototype for a potential system: an AI-powered waste sorter for hospital operating rooms. It takes the sorting decision away from staff in the middle of surgery.

My role
Co-founder, product creation and research
Course
ACAD 183 Product Management
Framework
Blue Ocean Strategy
Team
Six people, fall 2026
Triage, concept demo
Concept demo

The operating room is a waste hotspot, and most of it lands in the wrong bin.

Under time pressure, non-hazardous waste goes into the regulated "red bag." Hospitals then pay for specialized disposal of waste that never needed it.

20–30%of a hospital's total waste comes from operating rooms.
57–70%of OR waste is disposed of incorrectly.
~15%of hospital waste is actually hazardous (NLM).
6–8×higher disposal cost when waste is misclassified.

Avoiding solution jumping. My section.

I wrote and presented this part. Before the team picked any technology, we followed five rules.

  1. Start with the user.We looked at the problem from the OR circulating nurse's point of view: a sorting call made in seconds while focused on a patient.
  2. Ask "why" before "how."Staff aren't careless. Ambiguous items go in the red bag because it's the safe choice under regulatory risk and time pressure.
  3. Separate symptoms from the root cause.Overflowing red bags and high bills are symptoms. The root cause is cognitive load at the moment of disposal.
  4. Let the data define the problem.Mis-sorting rates of 57–70% showed where the real opportunity was.
  5. Challenge the industry's assumptions.Competitors ask how to haul waste better. We asked why it gets mis-sorted in the first place.

Training wears off and people turn over, so the fix has to remove the decision.

One container in. Five streams out, plus recycling.

  1. Drop inWaste goes into one container with several conveyor paths.
  2. IdentifyAn AI tracker classifies each item: infectious, hazardous, sharps, radioactive or general.
  3. RecoverThe sorter checks whether plastic or metal can be recycled. About 23% of surgical waste could be.
  4. RouteItems are routed into bins by category and recyclability.
  5. CollectPartners handle pickup, recycling and disposal.
Revenue
Monthly subscription plus maintenance instead of buying the machine outright.
Payback
The system pays for itself in disposal costs and fines the hospital no longer pays.
Partners
Blue-wrap recycling programs and haulers like Clean Management Environmental Group.

Incumbents optimize hauling. Triage removes the error.

Benchmarked against Stericycle, Stryker Sustainability Solutions and Practice Greenhealth, and automated sorters like MSW, Tomra, AMP and PICVISA.

Red ocean

"How do we haul waste better?"

  • Pickup and disposal after the fact
  • Relies on staff training
  • Results fade with turnover
  • Sorters built for municipal waste

Triage

"Why is waste mis-sorted in the first place?"

  • Works at the point of disposal
  • Takes the decision away from staff
  • Accurate on every shift
  • Recovers recyclables

A $366M U.S. market, starting with urban hospitals.

$12.96B
Total market: 215,977 hospitals worldwide at $60,000 a year.
$366M
Serviceable market: 6,100 U.S. hospitals.
$199.4M
Obtainable market: 3,324 urban U.S. hospitals.
$70M
The ask: build and lease 20 sorters, hire the team, run monthly accuracy testing, and reach the largest health systems.

Sources: Statista, "Number of Hospitals Worldwide" (2026); American Hospital Association, "Fast Facts on U.S. Hospitals" (2026).

Reflection and next steps

Framing drives the solution
The pitch got much stronger once we centered it on the nurse's split-second decision rather than on the machine.
Sequence is persuasion
The order you reveal information in is part of the argument.
Next steps
Validate with OR and environmental services staff, confirm recycling partners, and build a pro forma for the raise, starting with a single-site pilot.

What I owned on Triage

Problem framing
I wrote and presented "Avoiding Solution Jumping," taking the team from a symptom (overflowing red bags) to the root cause (cognitive load at the moment of disposal) before we chose any technology.
Visual Value Proposition
I created Triage's Visual Value Proposition: a one-page picture of the system with a single intake chute, an AI camera and a conveyor into six color-coded bins, plus callouts on how it works, a "today" panel on the 70% red-bag problem, the payoff numbers and our differentiator.
Story and motion
I timed the final deck's animations to the spoken narrative and built an animated web version of the pitch.
Stress-testing
I worked through the questions an OR nurse would ask, about sterility, sharps, sponge counts and turnover time, to find weak points before we presented.

For hospital operating rooms that overpay to dispose of ordinary waste because nurses must make sorting calls in seconds, Triage is an AI-powered waste sorting system that identifies and routes every item to the right stream automatically.

Team: Saissha Jain, Hamza Nassief, George Shannon, Jessy Elias, Keanna Bantay and Jackson Kaul.

02 / 06Venture conceptCo-founder

An online innovation hub where Iovine and Young Academy students build on intellectual property created at USC and take it toward market. The name stands for From Research to User.

My role
Co-founder
Context
IYA Innovator's Forum, Replit prompt
Stage
Prototype in progress, fall 2026
Outcome
Validated concept and pitch
fruit., concept demo
fruit. / IP library
  • Sample IP 01Biotech, patent
  • Sample IP 02Robotics, prototype
  • Sample IP 03Climate, concept
Team
Stage
IdeaValidationMarket
Concept validated
Concept demo

Four features take an IP from intake to market.

IP intake
Inventors submit their IP with a description, its status (patent, prototype or concept) and the fields it might fit.
Matching
IYA students browse the IPs, pitch an application and form a team around it.
Equity framework
A default split both sides agree to before work starts.
Progress tracking
Each pairing moves from idea to validation to market.

A validated concept, not a launched product.

The goal
Each team finishes with a validated business concept.
The finish line
A pitch to USC's Office of Research and Innovation, and optionally to the original researchers.
The handoff
Work is documented on the platform so later cohorts can continue it.
Open question
How equity is divided. Flagged and set aside for a later phase.

What I did on fruit.

Concept
I identified the two gaps in the planned class, IP sourcing and ownership, and shaped the hub around them.
Research
Grounded in an interview with the Dean of IYA, plus a student survey built on Mom Test principles.
Prototype
I'm building the hub prototype around the four features, to show a working version to the Dean.
Brand
I co-designed the fruit. identity, including the drip-"f" logo.
03 / 06AI softwareFounder

An AI agent that negotiates SaaS renewals on the buyer's behalf. Paste the renewal, get the counter.

My role
Founder
Type
AI agent for B2B software buyers
Built
Prototype, landing page, web app
Stack
Claude API, Next.js, Supabase
Contriq, concept demo
Contriq
Renewal notice

Your subscription renews next month.

Pricing will increase at renewal.

Reading termsBenchmarking, 20 vendors
Counter

We'd like to renew at current pricing.

In return, we can commit to a longer term.

Concept demo

From renewal email to counter-offer

  1. PasteThe buyer pastes in a renewal email or contract.
  2. ReadContriq reads the terms the vendor is proposing.
  3. BenchmarkIt compares them against a dataset I compiled for 20 major vendors.
  4. CounterIt drafts a counter-offer the buyer can send back.

Vendors in the benchmark

  • Salesforce
  • HubSpot
  • Datadog
  • Zendesk
  • Slack
  • Notion
  • Snowflake
  • Workday
  • DocuSign
  • Databricks
  • ServiceNow
  • Atlassian
  • Okta
  • Zoom
  • Twilio
  • Microsoft 365
  • GitHub Copilot
  • Adobe
  • Asana
  • monday.com

A working product, not a slide

Prototype
A working tool that reads a renewal and drafts the counter-offer, built on the Claude API and the benchmark dataset.
Landing page
A pitch page with a waitlist for early interest.
Web app
A Next.js and Supabase app with real authentication and database-level row security.
Name
"Counterparty," then "Backline," then Contriq after trademark checks. "Counter" stays in the product language.
Status
Paused for now. The next decision is the pricing model.
04 / 06Consumer AI appFounder

A roast-mode AI procrastination coach for college students. It checks in on your deadlines and gets less polite the longer you wait.

Taunt, concept demo
9:15Taunt
TauntEssay due at midnight. You have written 0 words.
TauntTwo hours later. Still 0. Bold strategy.
TauntExcuse court is in session. State your excuse.
My laptop died.
VerdictOverruled. The library has outlets.
Concept demo
Overview

At a glance

Excuse court
You plead your excuse for not starting, and Taunt rules on it.
Progress checks
Check-ins escalate when the answer is "nothing."
Knows college life
The roasts are specific to being a student.

The plan is to launch at USC and make Taunt something the whole campus talks about, while building it around the student so the business works beyond one school.

What I did on Taunt

Idea selection
I worked through and rejected admin utilities and a campus bounty marketplace before choosing the roast-mode coach.
Product definition
I chose the three core features: excuse court, progress checks and knowledge of college life.
Name and brand
I named the app Taunt and set up its brand system.
05 / 06Product conceptCo-founder

A thermal-camouflage suit that hides the wearer from infrared cameras.

WRAITH, concept demo
Suit offSuit onThermal camera view
Concept demo

Where WRAITH interrupts a thermal camera

  1. TargetBody heat is what reveals a person.
  2. OpticsThe reflective layer scatters infrared before the lens collects it.
  3. DetectorThe sensor picks up a weak, diffused reading.
  4. ReadoutWater cooling keeps the surface near ambient temperature.
  5. ProcessingWith no contrast, software has no silhouette to isolate.
  6. OutputThe image shows the background, not a person.

What I did on WRAITH

Product concept
I defined how the suit works: reflective shielding paired with water cooling.
Technical story
I wrote "How our design works," walking through the detection chain and where the suit breaks it.
Name and identity
I named the product and developed its logo: a hooded chevron crossed by a broken scan line.
Presentation
I helped build and deliver the Innovator's Forum presentation.
06 / 06Music data researchCo-lead

Does a TikTok trend cause chart success, or does it only amplify songs that were already rising?

My role
Co-lead, data model and visualizations
Data sources
TikTok, Spotify, Billboard, Chartmetric
Tracktion, sample data lag TikTok videosSpotify streamsHot 100
Concept demo

One song table, three platforms linked to it

Daily TikTok data, daily Spotify data and weekly Billboard data all link to a central Song table, and Chartmetric serves as the ID map that matches the same song across platforms.

lag_days
Days between a song's TikTok peak and its chart peak.
pre_trend_growth
Whether streams were already climbing before the trend. The cause-versus-amplify test.
weeks_after_peak
How long a song stays on the chart after its TikTok peak.
conversion_rate
Extra Spotify streams per 1,000 new TikTok videos.
is_catalog
Older catalog track or new release.

What I did on Tracktion

Co-lead
I co-led the project team.
Data model
I designed the Song table, the platform links and the five calculated fields.
Visualizations
I designed three mockups: a single-song timeline, holders versus spikers, and conversion by genre.