15 mn read

The Story of Drainpipe: How I built an AI Startup during a Global Pandemic

It’s January 2020, the week a helicopter crash rips away a living legend, Kobe Bryant, and the same week an opportunity I never dared to pursue opens itself up, inviting me to let it consume me. Today, we realize that dream. Where I am able to tell you the story of drainpipe: how I built an AI startup during a global pandemic. A platform to ingest any form of multimedia. Whether that was managing metadata, tagging things to know, and train AI to be able to detect those learnings. As well as, enhance awareness as data continues to be collected – at a scale and resolution not possible then. Still not yet possible today either. Here’s how it all started.

Background

While sitting next to a friend at a local bar in New York, I received a phone call announcing the end of what I had previously thought was my perfect career path. Until a moment earlier, I was a Director of Programmatic Trading at an NYC programmatic advertising company preparing for an IPO – What seemed like a perfect company and opportunity, suddenly wasn’t. And with one phone call, it was all over. In the sudden hollow of that moment, what I didn’t know at the time is that my friend beside me, Big Stevie, a former sand rock and gravel owner-operator whom we laughed “solid dirt to a graveyard”, is about to pave a path forward for me that reaches for a rainbow. He grabs my shoulder, like the uncle who knows the right move to part the cloudy skies. He asks what my dream is. When I answer he says, “Hey kid if you can do that, I’ll back ya!” He has one condition. What I build must be patentable. With a shot of B&B, Big Stevie yells stock orders into his flip phone and smacks his big fingers against his iPad, and when I’m ready to present, he listens, nods his head, and calls one of his advisors for another option. He’s on board.

Our Unique Approach

Here is what my experience was. I had previously owned and operated a Programmatic Advertising Company doing 4+ Billion transactions per day, and I had built and deployed software for banks and real-time routing systems for hazardous payloads. The processing pipeline video game assets follow inspired me. An engine usually streams the compressed format, which was fit for it, off of a disk into memory to be rendered. Years ago, that was a cartridge. Familiar with this topic as I was a graduate of Full Sail University in Orlando Florida. Directly spending time in Chicago programming slot games for one of the largest slot machine manufacturers in the world. I focused on using effects to make 2D images seem like 3D objects in memory-constrained environments. So, I get on the phone with Big Stevie’s people, thinking who could he possibly vet me with? This is an ace, nope. In fact, the guy I spoke with, Vance Turner, was a former systems admin of NASA’s Numerical Aerodynamic Simulator. I explained what the unique approach was at a high level and to my humble surprise he enthusiastically supported it. He reinforced that my thinking about the problem was correct. Going all in on this, I didn’t even take a side job. The first month and a half from the 3rd week in January until March 2020, was smooth. It was a lot of preparation, design, and research. It was easy to find Stevie because he used to call hi seat at the local bar “The Office.” That was only a few doors down from where I lived. But when COVID happened, it wasn’t so easy to see him.

From March Until November 2020

I worked based on our patent application. Having to think outside the box, I couldn’t just use an existing TensorFlow model, or what I would later catch on to be called a Foundational Model. I had to innovate extremely complex topics. Stevie and Dominick — Summer of 2020 The system to handle the feature points, was in part inspired by the world done at the Large Hadron Collider (LHC) at CERN. Work done in Neuroscience, partly inspired by high-resolution photogrammetry stitching research used in Radiology. As well as, loss-less compression techniques to get large amounts of parameters represented into a transferrable space capable of concurrent computation and distributed computing. Unlike what was existing at the time and even today. The system needed 2 components: the main web panel for users, and the computational mesh. Instead of choosing python, everything in the supply chain was done in either Javascript or business logic. Or, C/C++ for socket communication, converting multimedia at rest into features, and training AI models. Vance, having sat next to James Gosling at Sun Microsystems when he was developing Java. Harped that I manage my allocations and garbage collection for any performance-intensive jobs that we are scheduling to hardware. The reason is, for us to make the best use of the hardware, there could be no wasted clock cycles. The schedule had to be efficiently filled. That has been a big focus from the beginning.

Augmenting Social Data

Let’s augment social data. At the time Donald Trump was president, and the political atmosphere in America was boiling, running analysis on tweets was an interesting idea. Steve asked, “What can be patented?”. He required that there be patentable material from day one. We started digging in. I started writing code to prove it would work. As well as to understand the required process and architecture we need to patent. At the time I didn’t consider myself a system specialist or Artificial Intelligence Specialist. I had no money for food or rent. I was broke! Stevie began giving $200/week for food. We agreed on a number. I did enough work for the first draft of the patent by around March 2020 and started writing the architecture out. That’s when the Global Pandemic started.

Starting a business is hard. Coding a technology by yourself, to start a business is nearly impossible.

Beginnings of DrainPipe

When we started DrainPipe, I had no health insurance. My life had to be lean to get this accomplished. I had no car. So, I walked my groceries a mile from the store every few days. Living in a small town people would always see me and yell out of the window to ask if I wanted a ride. I always refused because I didn’t want to ask anything of anyone. It was a sacrifice to make, even when I was shivering in freezing temperatures. The pandemic complicated things further. I couldn’t see Stevie as easily. So, I walked in all kinds of weather conditions to see him. Whether that was rain, sleet, or frigid blizzards. A lot of the time to see him and talk to him. I’d make it to Stevie’s house, he would ask me to get a taxi and I would decline the offer. He would tear up. I used the adversity in front of me as fuel and growled through every condition there was. Walking up steep hills on the North Shore of Long Island. Rolling through the mud was my right of passage to Shambala.

Chance of a Lifetime

He was taking a chance with me that now else would take. At a time, no less, when no one would dare take a chance. I fully understood the gravity of my responsibilities to this man. He rightfully had brief moments of reluctance, though he would always tell me “I have all the faith in the world, in you kid!” I kept pushing, sometimes drinking 2 pots of coffee a day and talking to whomever I could on the phone about what I was working on. So, I could bounce ideas around. Even though everyone thought I was crazy. People who I considered friends told Stevie I was never going to get it accomplished. Not understanding the pedigree of the person I am.

May 2020

By May 2020, we had a product. It was just as slow, and clunky as existing technologies, but it was far more intuitive to use. It still wasn’t acceptable to me, and to drive more value to the patent. I need to make some changes. I had to break the technology up into 2 pieces. Design a scalable architecture, and rethink how we use mathematics to process the data. Which will be trained into an AI model. Coming from a video game background, early on I had been thinking about data similar to a video game. You run around the world in a game. It shows pieces of the world as you move around and streams them in and out of memory fast. I started learning Particle Physics, Rocket Science, and Neuroscience from a computational algorithm standpoint. A lot of math issues in these fields end up intersecting. But I’m not a rocket scientist. I had read thousands of white papers, I had to see what worked. Understand why it worked. Then figure out how to accomplish the holy grail of compressing data in a format ready to be trained into an Explainable AI model. Doing it in a way that would enable concurrency.

This Started the Most Difficult Endeavor of My Life

This was no longer a simple investment to make a few bucks, this innovation had implications. For one, it was going to take more time and there cost more money. But my Angel Investor said, “take your time, do it right, you never want to have to do this again.” So, that’s what I did. No money for help, just time and hard work.

It’s beyond scary to venture into building something and feeling like you are building a pyramid alone, piece by piece.

Especially needing to make mathematical breakthroughs to progress. Feeling far from knowing when I’d be able to put a dollar in my account. But having already jumped off the cliff and dedicated to this. There was no turning back, only gripping in deeper and stepping through it. I tried everything. It was unrelenting. October 2020 came, and I figured it out and had everything ready. We submitted our first patent application. Now I had to focus on making the product come to life. In November I managed to get drainpipe.io’s website live. I had to figure out how to store the data that the drainpipe produced. An easy task if looked at it short-sighted but when expecting to be able to sign people up and make the process easy, we had a plan for scalability.

So the new year came. It was 2021 

After walking over 1,000 miles in 2020, I still didn’t know when my first paycheck was coming. I dug deep and began changing my environment. I needed new experiences to weigh my brain into being able to address problems from different influences. Just like a neural network. By April 2021, I had it working. But I was so unsure if I did it right, that I doubted myself. So I called Vance, and he said “Dominick, you are a good engineer, you write good code, and you are going to make computers dream one day.” I spent many months and weeks continuing to build out the architecture to make a system scale and be able to run distributed workloads directly on hardware. In august 202, after a few medications on the website and a few articles written on artificial intelligence, I presented a prestigious title as a top 50 AI CEO and Innovator of 2021. At the same time, I received an email from a group in France that asked if we would like to be involved in a new type of event dedicated to Artificial Intelligence. It would be in 2022 and would be the perfect way to showcase an innovative solution in the space. The group took a liking to me and the idea of drainpipe. We determined I would speak at the event where we would exhibit and showcase drainpipe for the world.

Now I Had a Time Crunch, It had to be done!

So, in September, we decided to look for an early/interim Chief Technology Officer and start building a team. Within 24 hours of creating a job posting on LinkedIn over 30+ professionals applied. That list would grow to over 300+ faster than I could blink. I started scheduling calls all day to talk to people. Within 2 days of interviewing, we found our CTO. World-class talent aligned around a common goal. In fact, some of the people were so impactful in the progress of technology, emergency safety, medical information management, and databases that we simply couldn’t refuse their interest in the research. We owed them the opportunity to contribute and continue their life’s work. So, we built a team of rockstar advisors as well! All of this initially took me by surprise. There were highly qualified people, interested in what we did. When I explained DrainPipe, they only became more interested. I was gaining traction. We started seeing a growing list of people waiting on the launch of DrainPipe. Our selected interim CTO, Felipe Carino, former Chief Architect at Teradata said “it’s a miracle!”. When describing the complexity and sheer amount of work that had been done. More specifically, how it was done. No one person has ever embarked on a technological expedition of this magnitude. With their back against the wall no less. To then deliver a product with such implications for the future of technology. Our advisors only further solidified that by explaining common constraints faced in their fields and sharing their excitement for technology like DrainPipe.

It Was Starting to Feel Real

See in the 1980s there was an AI scare that Japan was going to use its computation power and take over the world with AI. In 1979 there was a joint study done between Caltech, and Citibank’s advanced banking division, that study was rolled into a company called Teradata. The United States Government, wary of Japan’s Computation Supremacy ultimately chose the intrapreneurial laboratory that made up NCR, Bell Labs, and Teradata to develop a high-performance computing system, and the individual who was the Chief Architect of that lab was Felipe. Having written numerous chapters in the IEEE standards, and works on Petabyte Scale Multimedia Databases, he was the ultimate person to validate our work. Felipe was no stranger to the problems of scaling the largest computer platforms on earth as the former Chief Architect at Teradata, and former Executive Director of R&D at MySpace.

Felipe Carino had this to say about our progress in a report from 2021:

This shall prove a terse reference to potential investors, customers, and strategic partners concerning the current state of Drainpip.io. Deep Dive and technology secret sauce can be provided under NDA. They keys to this exciting early-stage opportunity are listed below. • Drainpipe.io — has developed a unique Artificial Intelligence (AI) system and model that is flexible and not beholden to any specific company, like Facebook, Google, and other major AI players. • Drainpipe.io — two major innovations and uniqueness are (i) the innovative use of mathematical models not used before in AI modeling that enable scalability, performance, and accuracy — to provide highly accurate predictions and inferences; and (ii) the ability to integrate heterogeneous and hierarchical models and data sets for accurate prediction. • Drainpipe.io — current implementation was designed and coded for scalability. The current system is implemented and is not just a simple proof-of-concept in a single large code base, but a modular implementation using state-of-the-art available technologies. This system (details with NDA) uses similar algorithms and techniques that made Teradata the pioneer in scalable and very large systems. I write this as someone who was in charge of innovation at Teradata, ran an innovation group at MySpace, and worked at several early-stage startups.  — 

September 2021-January 2022

From September 2021 until January 2022 Felipe continued to work with us. He helped me gain confidence after all this time, and with a bet of this magnitude, the technical decisions made were correct. That’s not something you can just assume, in computer science all implementations is a hypothesis until observed otherwise through testing and evaluation. Something which Mr. Carino was uniquely qualified to do for this product. Felipe transitioned to focus on some other studies, while I carried on. From February to April the focus was preparing for our first roadshow. Which, in fact, would be debuting drainpipe.io at the first Annual World AI Cannes Festival. Where on the first day, I was slated to vie my first keynote speech titled “Space, Time, and Information.”

WAICF – World AI Cannes Festival 2022

Our booth was amazing, it was next to Microsoft, and CCI Nice. Much regarded as the Silicon Valley of France. A lot of distributed computing research came from this region of the world over the last 50 years.

I attended the Meta Global AI Leadership dinner while in France. Where I was surrounded by highly esteemed technologists and researchers from around the world. It was quite an honor. While on this trip, I made friendships that I believe will last the rest of my life. At our booth, we met with some of the largest companies in the world. We provided product demonstrations where I had the honor of seeing people react and respond to drainpipe for the very first time.

It was an Emotional High

When I got home from France, there was a tremendous amount of demand from some of the largest companies in the world knocking on our door finally. A great feeling after such hard work. I arrived back in New York on a Friday afternoon, and immediately ran over to hug Stevie and tell him how it went! he surprised me with plans to have a BBQ that Sunday and discuss all the progress we made in France. Steve, our Angel Investor, was a person who throughout the years of building drainpipe, became like family to me. He brought body-aching laughter out of me, he helped prepare me for reaching far beyond my dreams. I spent a lot of time with him talking about life and family, and he was a mentor to me. He mentored me on how to live a life filled with love and compassion. He guided me through some of the most difficult work I’ll ever have to endure. As well as, helped me maintain my sanity while working on problems that drove some highly intellectual mathematicians. Who made my work possible, to suicide. The man gave me one of the greatest gifts of my life. Which was the opportunity to chase my dreams and the hope of one day reaching my potential.

On that Sunday, April 24th

Two days after coming from France. While BBQing more food than most of the company parties I’ve ever attended. Big Stevie passes away in my arms. There’s not a word I can come up with to describe what it is like to lose someone who does such a thing for you, in front of you. It took the wind right out of my sial. I had black and blues on my body for weeks in the effort I took to save his life. I went from the best feeling in the world coming home from France to suddenly feeling empty and lost. Even though so much was yet to come. I lost not only a best friend. a surrogate family member, a mentor, and an investor.

Suddenly I was standing against the wind.

In the months that followed, I had to work through emotional pain, while progressing with what we started because now more than ever, failure was not an option. I began looking at how we were going to keep drainpipe alive long enough to bring this innovation to market. Sure we had deal opportunities. But they were very early. Moreover, it would take a while to lay the foundation for how these deals would be done with some of the largest companies in the world. However, Venture Capitalists I spoke with didn’t like the risk, didn’t see the value, and thought we had competition. Or, they flat-out didn’t believe in my ability to make this happen. Funny enough, many of them invested millions in FTX. So, I buckled down on the technology and focused on bringing it to where it needed to be.

September 2022

After some years spent in New York, it was time to move on to Daytona Beach, FL where the UCF Volusia county incubator located in the U.S. Customs Terminal at Daytona Beach International Airpot has embraced my journey with open arms. In November, I was honored with the opportunity to become a contributor to the AI for Radiology Topic Group at the World Health Organization and International Telecommunications Union.

Bring us to Current Events

As I foresaw at the bar that faithful day, writing an idea on a napkin; AI has become mainstream. Not without its problems. Which I remain mindful of. I chose not to lead drainpipe to the forefront but to methodically work through them prioritizing principles over profit. The world, climate, technology, and events influenced by all three are rapidly evolving. We as a civilization need to be better prepared for the unseen, unspoken, and unprovoked. Economies can no longer afford to wait for catastrophic events to occur, we must prepare for them. Small towns and large cities can no longer wait for potential mass casualty events to occur. We must prepare for them. Schools throughout America can no longer hope someone doesn’t come running in with a weapon. We must prepare for it.  

We learn a lot From History, But for the First Time, We can Now Prepare for the Future

Today, I am pleased to introduce you to the drainpipe. A vision I had for the future where we trained AI by combining any form of multimedia, easily defining what that data is, and using synthetic environments to build multi-domain datasets which will revolutionize preparedness.

An Operating System for Artificial Intelligence.

Utilizing environments operators can build complex scenarios and generate visual data to represent them. Collecting valuable insights without the expense of it playing out in real life. With a platform for building complex relationship graphs of ontologies, labels that represent any form of structured, or unstructured data. With a suite of tools starting first with an Artifact Inbox for any unlabeled data to be brought to a user’s immediate attention to the view and label. That data then becomes a class. Which can be used in a variety of ways with DrainPipe. Whether that is training models, used in integrations, or powering synthetic environments. Classes can be used to represent any form or combination of multimedia. Drainpipe is proud to announce sponsoring the World AI Cannes Festival for now a second year in a row. Where we have an exciting event planned.

Day 1 of WAICF

On the first day of the event Feb 9th at 12:00-12:30 PM CET I will be providing a workshop titled “Using Synthetic Environments to Train Artificial Intelligence.” Where we will be discussing the problems of working with different kinds of data and wrangling it. We will also show you how to use synthetic environments to build data sets for what we haven’t yet experienced in real life. Moreover, we will focus the workshop on manufacturing. As well as, the impact synthetic environments can have on preparedness for faults, defects, and catastrophic downtime.

Day 2 of WAICF

On the second day of the World AI Cannes Festival 2023 Feb 10th at 2:50-3:15 PM CET there is a historic event titled “The future of technology and humanitarian relief.” Additionally, we will have a special guest Nalini Saxena, speaking with us about the role technology plays in humanitarian efforts. Additionally, she will speak about economic disasters. We will also be showcasing to the world using synthetic environments to train drones how to find humans buried under rubble after natural disasters. Be sure to join us on this journey, it’s a moment you don’t want to miss. If you haven’t registered, or you can’t make the trip to France be sure to register for the digital experience at: https://www.worldaicannes.com/fr/register?payment_promo_code=WAICF23-DRAINPIPE

See you there!

Dominick Romano Founder, Drainpipe.io  

Recommend0 recommendationsPublished in Technology & Innovation