Leon Kuperman, CTO at CAST AI, on cloud automation, GenAI adoption, and scaling developer velocity

In this episode of CTO2CTO, Lucas Hendrich, CTO at Forte Group, talks with Leon Kuperman, CTO of CAST AI, about the compounding impact of automation, AI, and optimization at scale. They explore how developer productivity, cloud infrastructure, and org culture are evolving in parallel — and what that means for the next decade of tech leadership.

Velocity and Automation

LH: What slows teams down the most?
LK: Velocity. Or rather, the lack of it. When you stop releasing code, you’re on a death march. That’s when tech debt builds up. Automation helps solve that. At CAST, we release hundreds of times a day — because our platform is designed for speed, safety, and scale.

AI Isn’t Replacing Engineers — It’s Amplifying Them

LH: How do you manage different attitudes toward GenAI inside your team?
LK: I’ve got engineers who are skeptical and some who are leaning in hard. The self-motivated ones using tools like Cursor and JetBrains are just moving faster. We don’t have to force it — performance management will surface the difference over time. It’s becoming a self-solving problem.

The Economics of AI: Tokenomics, GPU Scarcity, and the Cloud Trap

LH: Is enterprise AI adoption being limited by infrastructure?
LK: Yes, and pricing is the big blind spot. Cloud providers are offloading GPU risk onto customers by locking them into 3-year contracts. It’s like buying today’s hardware at tomorrow’s prices. We’re building tools to find available GPUs anywhere — so you don’t get trapped in the wrong cloud or locked into the wrong vendor.

What Real AI Adoption Looks Like

LH: How do you think about where to apply LLMs?
LK: Use the simplest solution for the problem. Graph heuristics work? Use them. But for zero-shot time series forecasting? Foundational models trained on large corpora like Datadog’s make a real difference. You don’t need a hammer for every job — but you need to know which tools can scale.

CAST AI’s Philosophy: Automate the Low-Level to Elevate the High-Level

LK: We started with Kubernetes because it has structured rules, but our mission is bigger. Any area of software that can be automated at the infrastructure level should be — so engineers can focus on solving real problems, not just plumbing.

Brazilian Jiu-Jitsu and Building Resilient Teams

LK: I’m a brown belt in BJJ, and it’s taught me something crucial: hard things create clarity. When you’re in a tough spot on the mat, you don’t think about anything else — you learn to breathe, to survive, to wait for leverage. The same is true in engineering. You need calm minds, not busy ones. And that means fewer meetings, more flow, and less fear.

Stay tuned for more conversations at the intersection of technology, leadership, and AI on CTO2CTO.