Where Rubber Meets the Road
AI Entrepreneurship is one of the last courses in the USF’s Master of Science in Data Science curriculum taught by Uri Schonfeld. This class aptly wraps the program by allowing students to showcase their learnings in the past year. The course has a capstone project where students form startup groups and go through the motions of building a startup — from discovering a problem and conducting customer interviews all the way to producing a minimal viable product (MVP) and pitching to investors.
The course covers very broad content, covering the business aspects as well as technical aspects of starting a business:
- Ideation: how to find a problem worth solving and how to distinguish good ideas from bad ideas.
- Customer validation: how to conduct customer interviews to validate your hypothesis.
- Frontend: how to build a user-facing application using CSS, HTML, Flask, etc.
- Infrastructure: how to design and deploy an application using APIs, Docker, and cloud services.
- Positioning: how to market your business and stand out among competitors.
- Startup metrics: what metrics are important to track in the early stage.
- Pitching and fundraising: how to pitch and raise funding.
Select startups were given the opportunity to pitch their ideas to judges — venture capitalists, seasoned technology executives and entrepreneurs, and startup advisors — listed at the end of the article. Here are the select startups that were given the opportunity to pitch:
Justice Data Collective [1st Place]
Website: https://justice-data-collaborative.webflow.io/
Deck: DDC Pitch Deck
Team members: Ian Duke, Caleb Hamblen, Lance Santerre, Nick Miller, Emily Yin
The Justice Data Collaborative (JDC) created a powerful tool that empowers litigators and incarcerated people to fight for justice in the broken US prison system. The Justice Data Collaborative provides up-to-date historical data and predictions concerning prison population and lockdowns for all federal prisons in the US. Ian got the inspiration from time collaborating with attorneys and investigators at the Public Defender Service for the District of Columbia. Prison lockdowns deprive incarcerated people of their basic rights such as outdoor time, visits, legal calls, and participation in rehabilitation programs. With the lockdown predictions provided by JDC, families and attorneys can better plan their visits and legal calls so that they’re less likely to be turned away. Attorneys can also use the historical data as supporting evidence that their client had been deprived of basic entitlements and contextualize why they may not have been able to attend rehabilitative programs.
Podscicle [2nd Place]
App: https://podsicle.vercel.app/
Deck: Podsicle
Demo: Podsicle — AI Generated Podcast
Team members: Bassim Eledath, Irene Garcia Montaya, Sonal Shad, Ireri Avila, Zoe Le
Podscicle is an AI-generated podcast on the latest AI and machine learning research. The Podscicle team recognized the pain point of AI/ML professionals to keep up with the exponential speed at which AI/ML research is moving in. As a result, Podscicle allows AI professionals to use their downtime to listen to AI-generated podcasts based on research papers of their choice. It uses Llama and Gemini to process papers’ abstract sections to come up with interview questions and the rest of the paper to answer the questions. Then it uses a text-to-speech API to generate voices. Here is a sample podcast generated by Podscicle on Facenet.
InstaMuse [3rd Place]
App: https://huggingface.co/spaces/witchEverly/InstaMuse-Caption-Generator
Deck: https://www.canva.com/design/DAGD7EUPGeI/mwIll_5oW6D3RZqJ_d9pQw/edit
Team members: Bhumika Srinivas, Cassie Richter, Seneth Waterman, Jessica Brungard, Aditya Nair
InstaMuse is an automated Instagram caption generator for creators and businesses that makes it easy to grow their online presence. When an Instagram user wants to post new content, they can hit a writer’s block or they may not know the best caption and hashtags for optimal engagement. InstaMuse solves this problem by providing a few creative caption suggestions as well as trending hashtags from user uploaded photos. It leverages BLIP (Bootstrapping Language-Image Pre-training) framework to analyze the image and Google’s Gemini to produce the caption.
Emotion Explorer [3rd Place]
App: https://emotion-explorers-web-service-bg44mx37ca-wn.a.run.app/
Deck: emotion_explorers_shared_presentation.pdf
Team members: Amadeo Cabanela, Ronel Solomon, Krit Poshakrishna, David Ramirez, Yi-Fang Tsai
Emotion Explorer is an educational app for people with neurodevelopmental disorders (NDD), such as autism spectrum disorder, ADHD, learning disability, intellectual disorder, and epilepsy. Based on customer interviews with neuroscientists, therapists, and patients, the Emotion Explorer team built a real-time web app that uses computer vision to help NDD patients better understand emotions and facial expressions, creating a fun environment for social connection, confidence, and belonging. Its Scenario Explorer gives a prompt and asks patients to make the socially normal facial expression appropriate for in varying situations. Its computer vision technology detects the emotion and shows whether it’s correct.
Pawsitive Care
Deck: Pawsitive Care Project
Team members: Yen-Hsin (Cindy) Fang, Yazhu Jiang, Dawei Pang, Yihan Cao, Claire Zhou
The Pawsitive Care team built a web app for virtual pet care. There are two features that the team showcased in its MVP. The first feature is a virtual pet care assistant that provides 24/7, reliable pet care guidance via a LLM chatbot named Dr. Whisker, trained on veterinary handbooks for dogs and cats. This allows users to ask questions about emergency or urgent circumstances in natural language and can trust the responses generated based on credible sources. The other feature is PetBreedID, a pet breed detection using an image leveraging computer vision technology.
We were very lucky to have the following judges give their time to judge and provide feedback to our students:
- Scott Shiao: Scott is a Principal at Goodwater Capital, a $3B global consumer-tech VC firm. He previously led product at Zynga for the Cards division, and was a buy-side / sell-side equity analyst after starting his career as a software engineer building algorithmic trading systems.
- Peter Milford: Dr. Peter Milford is a physicist by training. Recently he has been advising early stage startup companies and has also been an advisor with Skydeck Ventures, a Berkelely-associated startup accelerator. Previously he was CTO/VP Engineering at Eyefluence Inc. a venture backed startup developing eye interaction technology for AR/VR headsets, acquired by Google in 2016, where he continued similar work, till he left in 2018.
- Karén Gyulbudaghyan: Karén Gyulbudaghyan is a visionary leader in global innovation & entrepreneurship ecosystems, strategic advisor, and investor. With over 30 years of diverse experience — from governments and Big Four consulting to multinational corporations and startups — Karén has driven significant impact cross-geographies and cross-industries. He helped transform UC Berkeley’s SkyDeck into a top global university accelerator and led groundbreaking initiatives at Yahoo!. As the founder of Strategic Value Ventures, he has elevated over three dozen startups like Krisp and SuperAnnotate to international fame.
- Shawn P. Calhoun: Shawn is the Dean of university libraries at USF and is currently serving as a co-lead on USF’s Entrepreneurship and Innovation Initiative. He is also a member of President Fitzgerald’s USF Leadership Team and a USF alumni.
- Jonathan Cromwell: Johnathan Cromwell is an entrepreneurship and innovation professor at the University of San Francisco, where he serves as faculty lead for the undergraduate major and faculty director for a new innovation incubator at USF. He has seen more than 150 project pitches across 6 years of teaching at the university, and he advises dozens of students on new startup ideas. He’s hoping to provide constructive feedback and share resources that USF can offer to support these projects moving forward
