CSE Research Catalyzes Protopia AI Startup and Earns Gartner Mentions

''''

By Kimberley Clementi

What materializes at the nexus between cutting-edge research and entrepreneurship? An innovative startup called Protopia AI.

Protopia AI is the name UC San Diego’s Computer Science and Engineering Professor Hadi Esmaeilzadeh has given the company he co-founded with long-term friend, Eiman Ebrahimi.

The business partners came across protopia in the writings of futurist Kevin Kelly.

The term refers to a destination that is not utopia, not dystopia, but rather something in the middle where problems and solutions converge—and where practical, meaningful ideas emerge.  For Esmaeilzadeh and Ebrahimi, Protopia AI’s Chief Technology Officer and Chief Executive Officer respectively, the relevance of the word’s meaning to their new venture was impossible to miss.

The two leaders launched their startup to occupy a practical space – a protopia, if you will – at the vulnerable juncture between any data source and data analytics. Protopia AI’s software, based directly on research Esmaeilzadeh conducted at CSE, proposes a practical solution to the inherent risks of accessing plain source data to perform sophisticated machine learning predictions.           

“It’s literally a translation of a research project that I did in UC San Diego. We published in 2020. We incorporated the company in 2021. And then Gartner picked us up in 2022. So, it’s a very close, very short timeframe – from cutting-edge research in AI to commercialization and entrepreneurship,” said Esmaeilzadeh.

Gartner Mentions for Innovation

Protopia AI’s innovative concept and fast-track to market are precisely what brought the startup to the attention of Gartner, a technology research firm that assesses vendors and shares their findings with customers searching for new tech solutions. According to Ebrahimi, a fledgling startup like Protopia AI doesn’t usually get mentions from Gartner this early into operations.

“When you talk to companies that have been talking with Gartner, often they’ll tell you it takes them maybe a year, two years to get any mention,” said Ebrahimi. “We have been talking to Gartner for about six months, and in the past one month, there have been two pieces that have mentioned Protopia AI.”

Gartner initially recognized Protopia AI with a Cool Vendor Award which highlights vendors who are bringing unique and new technology to the market. Most recently, Gartner mentioned the startup in their Hype Cycle Report. This report defines expectations for new technologies across time, beginning with the innovation trigger through maturation.

While Esmaeilzadeh and Ebrahimi are honored by the two mentions from Gartner, they are deeply excited about the implications for Protopia AI. Gartner’s Hype Cycle Report suggests the company’s timeline to productivity could be as short as two to five years. This projection comes at an opportune moment – just as the startup has closed a round of funding and is accelerating work with their first few marquee customers.

Pioneering the Problem

For Esmaeilzadeh, this particular journey began when he ventured outside his immediate scope of expertise as a computer architect and started a line of research on privacy preserving AI. The resulting paper, Shredder: Learning Noise Distributions to Protect Inference Privacy, addressed the privacy issues surrounding AI analytics while maintaining inference accuracy.

“We are at an inflection point when it comes to AI and machine learning. We have finally reached the point that we are actively using these services around our environment,” noted Esmaeilzadeh, referring to the explosion of IoT (Internet of Things). “They are very much embedded in our homes, in our offices, in our social media. And they’re literally preying on our data.” 

The complexities surrounding data privacy stem from the nature of the data in storage as well as its sheer volume. Stored data includes our images, voice recordings and, in some cases, very private and sensitive information. Compounding the problem, data collection is no longer byte size but rather measured in exabytes. Currently, all that raw data could be vulnerable when it is utilized for vital services.    

 

“As individuals, we’ll use public transport or go to restaurants or other public venues where video of the environment is collected and used for analytics. Often the analytics comes from a third-party provider with no guarantees about how the data may be exposed in the process,” explained Ebrahimi. “The issue is that the entities that capture our data are responsible for it, but then pass our data, including sensitive information, to providers where privacy is not a priority.”

Current practices like these raise an urgent question: How can those responsible for our data gain useful insights from it without infringing on our privacy or exposing our most sensitive information to unintended parties when it is accessed for analytics? Esmaeilzadeh and Ebrahimi believe Protopia AI is the answer.
 

Pioneering the Solution

When Esmaeilzadeh initiated his UC San Diego research project into privacy preserving AI, his aim was to explore the issues surrounding data ownership. He had no intention of launching a startup. As the project developed, Esmaeilzadeh devised a method to protect private and privileged data sent from any data source.

Protopia AI’s software, dubbed Stained Glass, garbles each individual data record using an innovative algorithm. These transformations enable deep learning models to consume the data and make accurate predictions without exposing the original raw information in the process. Instead, the data is reduced to the minimal form necessary to accomplish the task before it is transferred to complete analytics. This means the most sensitive data never leaves the initial storage device or sensor.

While data security is paramount, the leadership team believes their software has achieved another important objective – a more ethical way of doing AI. Vital services can be provided without any tells in the data to reveal gender, race or other unrelated demographics.

“The question is not, ‘Why should anyone do it?’ The question really becomes, ‘Why wouldn’t everyone want to do it?’” said Ebrahimi.

Protopia AI joins a list of nearly 60 ventures founded or co-founded by CSE students, graduates and faculty. UC San Diego’s CSE department prides itself on an entrepreneurial spirit to find solutions to the world’s problems.