Our Team

Mohit Gupta

Co-founder, Opas AI
Mohit has spent the last decade working on wide variety of interesting problems in Infrastructure, Big Data, and ML. He has run and managed infrastructure at massive scale and felt the pain of operating them first hand. This pain combined with his experience in Big Data and ML led to the incubation of Opas AI. Before founding Opas AI, Mohit has held multiple roles at BloomReach, Groupon, Google, and Amazon. He has a Bachelor in CS from IIT Kharagpur and Masters from UCDavis.

Gurashish Brar

Co-founder, Opas AI
Prior to founding Opas AI, Gurashish led a team of engineers at BloomReach that built big data infrastructure spanning the globe guaranteeing low latencies and high availability in an inherently unreliable environment. Before this, at Microsoft, he worked to ensure the world stayed on time with Outlook Calendar and developed a revolutionary application streaming technology that delivered Microsoft Office on demand to consumers and formed the backbone of Office 365. He has Masters from Georgia Tech and Bachelor in CS from IIT Kharagpur.
Opas is based on the frustrating reality that our founders, Gurashish and Mohit, encountered in their daily professional lives: Today’s applications and infrastructure have evolved far faster than the technology used to measure and manage them. Watching their IT teams drown under the weight of too many tools, too many dashboards and too many meaningless alerts, they decided to do something about it.
The vision behind Opas was bold:  to create a single solution that can find, diagnose and fix even the most complex performance and availability problems in your applications and infrastructure. With that vision in mind, Mohit and Gurashish set about finding the world-class talent in data science, AI and ML needed to develop this solution.

Today’s application infrastructure is increasingly complex, but ML algorithms capable of understanding  them are data-hungry. Monitored data on the other hand is primarily human-focused and is incapable of satisfying the ML needs. Focusing on this data problem, the team’s first hire was Dinakar Dhurjati, who had been at Qualcomm building a web application firewall that utilized machine learning  to detect zero day attacks. His experience in applied machine learning along with formal and Domain specific languages led him to identify the inherent inefficiencies in how traditional monitoring systems collect data and develop a new design that learns the behavior of the system and can monitor 100 X more metrics while matching the overhead of traditional collectors.

Since no human can visualize or understand this volume of metrics to get to the root cause of a problem, we shifted our focus to building algorithms capable of learning on their own to identify the root cause of any performance problem. Vinesh Gudla, with his past experience building bayesian models for generating recommendations in the relevance team at Groupon and churn prediction models at datacoral, was our next hire. Vinesh began by building a basic statistical framework to understand the nature of the time series data; leveraging information theory, he brought out relationships between metrics, clustering them based on their shared distributions. The next step was to understand each metric on its own to see what is normal and abnormal, all the while keeping in mind the expenses associated with RNNs.

To identify the root cause driving the anomalous behavior, Vinesh built our first version of real-time root cause analysis as a multi-stage pipeline called Entropy Constrained Correlations. Leveraging information theory and using the inferred topology as strong constraints, he was able to deliver promising results where we are able to narrow the answer down to 10-20 metrics with a high probability that the true root cause is captured.

When you’re building a world-changing product, you need a world-class team. The core Opas team are Graduates of some of the world’s most innovative universities. Professionally, they have leveraged their experience in machine learning and AI as key players at companies that have changed the way we work, live and think. Now they’re applying this expertise to solve the toughest application performance problems.

But correlation is not causation. Pablo Jadzinsky spent most of his career studying time-series analysis and he knew the fallacies of correlation. Pablo did a Ph.D. building protein models with Roger Kronberg (2006 Chemistry Nobel prize winner) and then moved on to studying how visual signals propagate and are processed through the nervous system and later at Vium building models of how behavioral metrics can be used to detect disease. His first question was whether customers could be abstracted, using data from one customer to learn the behavior across all of them; he then built causation models that will keep improving as Opas grows and more data flows in.

While algorithms and data are intriguing for data scientists, we knew that our users are overburdened who often have to respond in the middle of the night to performance issues. To ensure the optimal user experience, we recruited Julian Wong, a veteran front-end developer and UX design with extensive experience building front-end applications at BloomReach and Yahoo. His code has literally touched the lives of hundreds of millions of people across the globe. To him, every line of code written is a work of art that has to convey the cryptic binary numbers being streamed through REST API into a beautiful picture that not only appeals to the user but gets her the answer she seeks with as little effort as possible.

 

The journey at Opas is just starting for all of us. We have built unsupervised algorithms capable of learning causal relationships, understanding the abnormal behavior of large distributed systems and identifying root cause of any performance problem in real time. The challenge is to expand our technology stack footprint and customer base and ensure our algorithms get better with each new customer, ultimately evolving our causation models to predict outages before they happen.

Board of Directors

Mohit Gupta

Co-founder, Opas AI
Mohit has spent the last decade working on wide variety of interesting problems in Infrastructure, Big Data, and ML. He has run and managed infrastructure at massive scale and felt the pain of operating them first hand. This pain combined with his experience in Big Data and ML led to the incubation of Opas AI. Before founding Opas AI, Mohit has held multiple roles at BloomReach, Groupon, Google, and Amazon. He has a Bachelor in CS from IIT Kharagpur and Masters from UCDavis.

Gurashish Brar

Co-founder, Opas AI
Prior to founding Opas AI, Gurashish led a team of engineers at BloomReach that built big data infrastructure spanning the globe guaranteeing low latencies and high availability in an inherently unreliable environment. Before this, at Microsoft, he worked to ensure the world stayed on time with Outlook Calendar and developed a revolutionary application streaming technology that delivered Microsoft Office on demand to consumers and formed the backbone of Office 365. He has Masters from Georgia Tech and Bachelor in CS from IIT Kharagpur.

Ashu Garg

General Partner, Foundation Capital
Ashu Garg is a General Partner at Foundation Capital where he invests in enterprise software – in particular, marketing technology, cloud infrastructure, and data platforms. He currently serves on the boards of QuanticMind, Localytics, Conviva, ZeroStack, Fortanix, Opas.ai, and Custora among others. Ashu also led our investments in Tubi and Cohesity. Ashu led Foundation Capital’s investments in TubeMogul, which went public and was subsequently acquired by Adobe; Aggregate Knowledge, which was acquired by Neustar; and FreeWheel, which was acquired by Comcast. In addition, Ashu authored the definitive whitepaper on how technology will transform marketing, MarTech and the Decade of the CMO.

Jon Gelsey

CEO, Xnor.ai / Ex-CEO, Auth0
Jon Gelsey is CEO of Xnor.ai, a computer vision and ML spinoff of the Allen Institute for AI and UW funded by Madrona.  Before Xnor, he was CEO of Auth0, an industry-leading identity-as-a-service platform, which he grew from 5 employees to nearly 300 over four years.  His previous experience includes responsibility for acquisitions and investments at Microsoft and being a venture investor at Intel Capital.  Jon started his career as a supercomputer designer at Convex Computer, acquired by Hewlett-Packard in 1995 as the high end of their server product line.

Pin It on Pinterest