Razi Raziuddin, Co-Founder and CEO of FeatureByte – Interview Series

Razi Raziuddin is the co-founder and CEO of FeatureByte, with a vision to unlock the last hurdle to scaling AI in the enterprise. Razi’s analytics and growth experience includes leading teams at two unicorn startups. Raz helped DataRobot grow from 10 to 850 employees in less than six years. He pioneered the service-led go-to-market strategy that has become a hallmark of DataRobot’s rapid growth.

FeatureByte is on a mission to advance enterprise AI by radically simplifying and industrializing data. The Functional Engineering and Management (FEM) platform enables data scientists to create and share modern features and production-ready data pipelines in minutes – instead of weeks or months.

What initially attracted you to computer science and machine learning?

As someone who started coding in high school, I was fascinated by a machine that I could “talk” to and control with code. I was instantly hooked on the endless possibilities of new applications. Machine learning represented a paradigm shift in programming, allowing machines to learn and perform tasks without even being instructed to follow steps of code. The limitless potential of ML applications is something that excites me every day.

You were the first business hire at DataRobot, an automated machine learning platform that enables organizations to become AI-centric. Then you helped grow the company from 10 to 1,000 employees in less than 6 years. What were some of the key takeaways from this experience?

Going from zero to one is difficult, but incredibly exciting and rewarding. Each stage of a company’s evolution presents different challenges, but seeing a company grow and succeed is an amazing feeling.

My experience with AutoML opened my eyes to the unlimited potential of AI. It is exciting to see how this technology can be used in many different industries and applications. At the end of the day, creating a new category is rare but incredibly rewarding. My key takeaways from the experience:

• Create an amazing product and avoid chasing fads
• Don’t be afraid to be the opposite
• Focus on solving customer problems and delivering value
• Always be open to innovation and trying new things
• Create and implement the right company culture from the ground up

Can you share the origin story of FeatureByte?

It’s a well-known fact in the AI/ML world that big AI starts with big data. But preparing, implementing and managing AI data (or features) is complex and time-consuming. My co-founder, Xavier Konort, and I saw this problem firsthand at DataRobot. Although modeling has been greatly simplified thanks to AutoML tools, feature engineering and management remains a formidable challenge. Based on our combined experience and expertise, Xavier and I felt we could really help organizations solve this challenge and deliver on the promise of AI everywhere.

Feature engineering is at the core of FeatureByte, can you explain what that is for our readers?

Ultimately, data quality drives the quality and performance of AI models. The data that is fed into the models to train them and predict future results are called features. Features represent information about entities and events, such as customer demographic or psychographic data, or the distance between the cardholder and the merchant for a credit card transaction, or the number of different categories of items from a store purchase.

The process of transforming raw data into features—for training ML models and predicting future results—is called feature engineering.

Why is feature engineering one of the most challenging aspects of machine learning projects?

Feature engineering is very important because the process is directly responsible for the performance of ML models. Good feature engineering requires three fairly independent skills to come together—domain knowledge, data science, and data engineering. Domain knowledge helps data scientists determine what signals to extract from data for a particular problem or use case. You need data science skills to extract these signals. Finally, data engineering helps you lay out the pipelines and perform all these operations at scale with large volumes of data.

In the vast majority of organizations, these skills reside in different teams. These teams use different tools and do not communicate well with each other. This causes a lot of friction in the process and slows it down to stop grinding.

Can you share some insight into why feature engineering is the weakest link in AI scalability?

According to Andrew Ng, a well-known artificial intelligence expert, “Applied machine learning is basically artistic engineering.” Despite its criticality to the machine learning lifecycle, functional engineering remains complex, time-consuming, and dependent on expert knowledge. There is a serious lack of tools to make the process easier, faster and more industrial. The effort and expertise required prevent enterprises from being able to deploy AI at scale.

Can you share some of the challenges behind building a data-driven AI solution that radically simplifies feature engineering for data scientists?

Creating a product that has a 10x advantage over the status quo is very difficult. Fortunately, Xavier has deep data science expertise that he leverages to rethink the entire feature workflow from first principles. We have a full team of world-class data scientists and engineers who can make our vision a reality. And let users and development partners advise on UX simplification to best solve their challenges.

How will the FeatureByte platform accelerate data preparation for machine learning applications?

Data preparation for ML is an iterative process that relies on rapid experimentation. The open source FeatureByte SDK is a declarative framework for building cutting-edge features with just a few lines of code and deploying data pipelines in minutes instead of weeks or months. This allows data scientists to focus on creative problem solving and iterating on live data quickly rather than worrying about plumbing.

The result is not only faster data preparation and service in production, but also improved model performance thanks to powerful features.

Can you discuss how the FeatureByte platform can further simplify various ongoing management tasks?

The FeatureByte platform is designed to manage the end-to-end ML feature lifecycle. The declarative framework allows FeatureByte to automatically deploy data pipelines while extracting metadata relevant to managing the overall environment. Users can monitor pipeline health and costs and manage feature line, version and correctness from the same GUI. Enterprise-class role-based access and approval workflows ensure data privacy and security while preventing feature proliferation.

Is there anything else you’d like to share about FeatureByte?

Most enterprise AI tools focus on improving machine learning models. We set out on a mission to help enterprises scale their AI by simplifying and industrializing AI data. At FeatureByte, we’re addressing the biggest challenge for AI practitioners: to provide a consistent, scalable way to prepare, serve, and manage data throughout a model’s lifecycle, while radically simplifying the entire process.

If you’re a data scientist or engineer interested in staying on the cutting edge of data science, we encourage you to experience the power of FeatureByte for free.

Thanks for the great interview, readers who want to learn more should visit FeatureByte.