n this interview, we speak with Lucas McGrew, Sumatra.ai’s CTO and Co-Founder leading our engineering team to build a scalable and accessible solution to providing realtime context for AI. Lucas shares his insights and vision for the company, highlighting its genesis and the drive to make AI accessible to all and what we have prioritized in building out the architecture for the platform. 

What led you to decide to build Sumatra?

I spent a lot of time building decisioning systems and realized the power of enriching real-time data with models to make inferences. The ability to automate interventions at scale was remarkable. I wanted to bring this capability to more people, but it was a hard problem that required getting many things right. Initially, it was only accessible to well-leveraged companies. I saw how algorithmic trading took over market trading, eliminating manual traders, and how similar approaches were used in platforms like TikTok and Instagram to increase engagement. I wanted to apply the same techniques to unlock all kinds of possibilities for real-time data to create seamless experiences.

What early decisions in building Sumatra do you feel are paying off today?

We made early investments in stability, performance, and security. We chose an AWS serverless stack for deployment, operational needs, and monitoring. This decision has allowed us to do more with fewer resources and stay pretty lean without overburdening the team. 

What is the core value of Sumatra that you are most passionate about?

I am most passionate about enabling individuals to leverage and scale their specific domain expertise. Software engineering is the application of someone else's domain expertise. We aim to enable people to do this at scale, whether it's recommending products and content based on user engagement or screening out bad actors and bots in real time.

What part of your role do you enjoy the most and why?

Personally, I find great satisfaction in building things and solving architectural and scale problems. It's incredibly rewarding to tackle challenges head-on and witness the solutions come to life. However, what truly motivates me, and our team as a whole, is the opportunity to democratize the ability to build complex models that were previously out of reach due to these very problems. We are driven by the belief that everyone should have access to powerful tools and technologies that enable them to innovate and create. 

One of our recent releases was focused on just this – solving one of the biggest pain points of deploying an ML model—waiting for an engineer to spin up a new dedicated service (e.g. Kubernetes) for the model. Now anyone can deploy a new serverless model themselves within Sumatra, so there's no back-and-forth between teams. Users can view and manage models directly in the Sumatra UI so you can easily track all of your models and model versions.

What has been the biggest challenge in building Sumatra?

The biggest challenge has been bringing the power of our technology to users without requiring them to build up extensive training and knowledge. We focus on effective abstractions and user experience, ensuring that users can be immediately productive and have a gentle learning curve while adding value incrementally. Simplifying the onboarding experience and identifying the people who can benefit the most from our immediate value are crucial. We’ve been really focused on that first-mile experience for users as a foundational principle of our product. 

What work in your career are you most proud of?

I am most proud of building Sumatra and being a founder. I've always been part of small teams, but this is my first experience as a startup founder. It's been an exciting journey, from finding product-market fit to marketing and everything in between. Similar to identifying trading strategies, we experiment and adapt to the market as a whole.

Why was Greg Kuhlmann the right partner to build Sumatra with?

Greg was the right partner because he has a complementary background in data science. He knows where to apply our technology in novel and innovative ways, understanding the short-term problems and long-term vision. He also shares a passion for building culture and teams, which has been invaluable.

How do you see the partnership between engineering and data teams evolving?

In the long term, I see a merging of engineering and data teams as better abstractions enable more people to perform tasks without specialized training and knowledge. We aim for a more collaborative approach to requirements and solutions, embracing an iterative and constantly changing process. The interaction between end product, product data, engineering, and data teams will increase, especially in the realm of real-time operations. Customization and personalization can be done without constant engineering intervention. The engineering team sets up the core architecture and competency, allowing for trust in the process and empowering faster remediation.

Learn more from Lucas

• Connect with Lucas on Linkedin 

• Check out our release notes to hear more about what he and the team have been working on.