3 broke high schoolers built a business impressive enough to present at Harvard’s Infamous Research Conference - here’s how they did it
The Source:
Archit Thanikella, avid car enthusiast, consistently kept his eyes on the roads in his community. Soon enough, he identified a significant problem - potholes causing severe car accidents - a problem he knew he could fix. His first solution was an anomaly detection software called Fovea, which helped drivers detect where potholes were in real time to avoid accidents. Archit pitched this idea to Siddhant Singh—a programmer and serial entrepreneur. Siddhant, after visiting India over the past summer, realized declining road infrastructure was a huge problem with no real solution - and hopped on board. At the same time, Fovea had a separate project in development - software that could identify infrastructure damage for municipalities. They were able to take this idea to Genius Olympiad - where they placed 1st out of 1600 submissions for their category. At the same time, Paul Joseph, a separate entrepreneur at this time- had been able to gain direct mentorship under his city’s chamber of commerce to conduct research on road-related issues - to try to gather more user research. This led to his publication into the International Conference on Frontiers of Artificial Intelligence and Machine Learning (FAIML IEE) conference along with being recognized as a Young Hero by TD Ameritrade. While his pathway wasn’t as clear, Paul knew that he wanted to make his efforts more known to raise awareness for the steady increase in car accidents. By raising $1.1k in grant funding, he was able to contact Brian Moen, the city director of Frisco, CA, to discuss the next steps and implementation. Due to a clear overlap in infrastructure related projects, Archit and Siddhant were able to team up with Paul for further development of their ideas. Fovea would eventually become an idea of the past as market validation revealed the software wasn’t implementable in the market.
The Transformation:
Over time, a key connection would allow for these three to progress from their pothole detection software to something greater. A cloud security engineer at DataDog, a major company in the industry, would agree to meet with Paul, Siddhant, and Archit. While their initial ideas seemed to be unfeasible, they would enter the meeting defeated - with no ideas on where to go. As they arrived in the conference room with the engineer, their initial pitch deck emphasized trying to communicate data between different semi-autonomous vehicles to help avoid crashes. The engineer shut the idea down due to the liability issues - if someone used their software and got into a crash, Archit, Paul, and Sidddhant could be sued for millions of dollars. Rather, he suggested focusing on “factory system optimization,” and trying to make existing processes more efficient As the engineer elaborated on this concept, a strategy to break away from their previous idea had emerged. Near this time, one of Archit’s friends had fainted while working in a factory, resulting in a minor concussion; making the three entrepreneurs ever more aware of the stress that strenuous labor put on workers over time. They had finally decided on a problem. Through a B2B model (selling to companies rather than consumers), their idea would be known as NOVA, a dashboard application providing analysis of the issues that pervade supply chain companies in manufacturing processes, as well as feedback on how to improve their efficiency. NOVA would employ a SaaS model (Software-as-a-Service) that used indicators like the number of machines, time-used, and resource intensity to create a customized, cost-friendly solution for clients. Through a monthly tier based subscription, NOVA was able to utilize APM (application performance-monitoring), fleet profiler, log management, various anomaly analysis features that employ K-Means Logistic Regression techniques, and novel anomaly detection through NOVA’s CNN architecture in order to create a sustainable solution. Along with that, slow improvements in the development of the dashboard configuration would allow for the development of new features like custom alert systems, detailed machine performance tracking, and custom mapping for existing factory equipment. What started as a pothole detection software had transformed into a customizable tool that different factory systems could use in their manufacturing assembly lines to decrease unplanned downtime - and eventually save lives.
Figure 1: Prototype of NOVA’s anomaly detection software in usage
NOVA Now:
How can baseline activity within a factory system vary from entity to entity? Ambitious attempts at compiling large amounts of data streams can lead to higher risk of data breaches which can reduce the trust and reliability of the company. The most apparent threat is disruptions within the economy as the supply chain is directly reliant on the basic economic principle of supply and demand. In terms of recession, businesses have lower demand for supply chain management services due to large portions of their capital going towards keeping a positive profit margin. During times of economic ruin, clients who go out of business or have to scale down their operations lead to an influx of data that can be analyzed by anomaly detection softwares between competitors.
Increased accuracy rates of anomaly detection models can lead to a phenomenon known as predictive maintenance where sensor data from machinery within factories can be used to reduce downtime and schedule repairs. ADS can also be used in combination with environmental sensors within a manufacturing factory or industrial plant where air quality, humidity, and temperature can directly affect the machinery. As they continue to look for partnerships, NOVA can serve as a personalized approach for factory systems that look outdated every single day due to higher turnover rates. NOVA’s business model follows a subscription based approach in providing accurate cloud-based monitoring and analytics platforms. This business model consists of a 3 tier level for a technology license, NOVA’s services, and various software extensions. The free version allows for pilot test runs where companies can test out NOVA’s services and consists of 5 basic features with less time to analyze the data collected on data processes within a factory system. The pro feature is $12 per host and is a collection of 12 features that are directly related to anomaly detection where multiple data points are compared to predict the next downtime. The enterprise distribution is listed at $21 per host and consists of 20 features that have detailed machine learning analytics which can be used in unison with the anomaly detection models. Some of the clear licenses include the ability to monitor the overall performance/health of factory equipment and being able to work with IoT devices which isn’t common in this market. NOVA is able to offer its services through a paid subscription which is a variable cost, as the amount of customers are not constant. Some software extensions include options such as monitoring AWS Fargate, customer chat features, and sound features for the alert systems. Regardless of NOVA being in its initial testing stages, there have been potential clients that specialize in the additive manufacturing department such as Wahlen Tech Pvt. Ltd. This potential customer specialties the creation of transmission and engine systems for automotive industries. A company that has been able to optimize similar models that NOVA proposes is TIBCO Spotfire Automation Services software. TIBCO Spotfire Automation Services software has been able to help Hemlock Semiconductor (HSC) through its anomaly detection to control their semiconductor manufacturing. Through generated alerts in times when factory machinery falls out of parameter bands, the specific anomalies can be detected in order to prevent downtime from happening in an essential industry. This service provided by TIBCO has allowed for HSC to save nearly $300,000 monthly which showcases how companies specialized in anomaly detection can predict downtimes at an efficient rate while specific modalities can allow for predictive maintenance. NOVA’s feasibility primarily is shown through its modalities which are offered as additional features on top of the standardized dashboard platform allowing for companies to gain further versatility in how their manufacturing processes are controlled.
Figure 2: NOVA’s B2B (Business to Business) Structure
Figure 2: Model of Anomaly Detection Through Rotor Speeds (one of many facets of NOVA’s software)
NOVA Tomorrow:
Moving into 2023, these 3 hope to compete in a wide variety of accelerators, business competitions, and receive funding for their idea which can allow for a physical prototype to be developed. By receiving feedback, NOVA hopes to secure further partnerships with manufacturing factories that can make use of this software to protect the health of their factory workers. While NOVA has considered a variety of methods in order to find the appropriate client, a horizontal SaaS (selling to a wide audience of businesses regardless of their industry) approach means that their services can offer more versatility than competitors, giving them an advantage. Specific needs of these industries vary but clear companies fall under the retail, logistics, healthcare and manufacturing sectors. NOVA will further offer APIs to be connected with other successful platforms in order to prevent downtime in the most efficient way possible. Certain parameters will be preset while others will be implemented pertaining to the requests of the client. Premier examples include inventory management, along with sourcing for equipment and systems which is indirectly done through NOVA’s systems.
Figure 4: NOVA’s Implementation into Supply Chain Processes at Retail Stores
Figure 5: NOVA’s organizational structure
The Outlook
Now at the crux of this long and hard journey, it dawns on our team that we have completed one milestone of many, and that we have such a long way to go. This is the plight of any startup, and rather than looking up a mountain with disdain and reluctancy, we take it one step at a time, putting one foot in front of the other, and occasionally looking back to appreciate our growth - both in success and failure. Our entire team wants to thank Arjun and the Exemplar Journal Team for providing us with this amazing opportunity, and it was an absolute blast to work with such a group of professionals while bringing this article to fruition. For any entrepreneurs with ideas that have huge potential but simply don’t have the exposure that they deserve, or for a struggling entrepreneur in need of motivation or a helping hand - please reach out to Arjun - sit down and talk to him for 15 or 20 minutes. I have never in my entire life felt more understood than during Arjun and our teams first meet, and this team has benefited so much because of it. The Exemplar Journal is a truly special venture and platform, and it’s bigger than any one person. Entrepreneurs around the world - together, we can make our planet a better place.
Author : Arjun Premnath
Founder and CEO of Exemplar L.L.C and the Exemplar Learnings Foundation
Arjun is a management student at the Kelley School of Business who is passionate about business, entrepreneurship, and finance.
In his free time, he loves to read books and work on his ventures.
Co-Authors : Paul Joseph, Archit Thanikella, Paul Joseph
NOVA’s Organizational Team