What if the real future of AI in engineering isn’t flashy demos—but boring, repeatable automation?
In this episode of CTO2CTO, Forte Group’s CTO Lucas Hendrich sits down with Leon Kuperman, CTO at CAST AI, to talk about building developer velocity, rethinking infrastructure, and why leadership in the AI era requires letting go of control to create more flow.
It’s a conversation that cuts through hype and goes deep on the pragmatic realities of leading tech teams today.
CAST AI’s entire model is built around reducing cloud complexity and maximizing performance through automation.
For Leon, that same principle applies to teams.
“When you're not releasing code, you're on a death march. That’s when fear creeps in and tech debt piles up.”
By automating infrastructure—down to real-time optimization of Kubernetes clusters—his team pushes code hundreds of times per day. The key: reduce decision friction, eliminate unnecessary steps, and create space for engineers to focus on real problems.
Leon doesn’t sugarcoat it: some engineers are skeptical about GenAI. But inside CAST AI, the ones leaning into tools like Cursor or JetBrains AI are shipping faster, solving harder problems—and separating themselves in the process.
“We don’t mandate it. The performance gap is making this a self-solving problem.”
This is the real shift: GenAI isn’t about replacing developers. It’s about amplifying those who know how to wield it well.
AI for AI’s sake? Not at CAST AI.
Leon emphasizes using the simplest solution for the problem—whether that’s graph heuristics or a large foundational model. He shared how CAST AI uses LLMs for very specific use cases (like zero-shot time series forecasting), and resists the temptation to over-engineer.
“You don’t need a hammer for every job—you need to know which tools actually scale.”
According to Leon, the biggest blocker for AI adoption isn’t just tooling or training—it’s infrastructure pricing.
Cloud vendors are shifting GPU risk to customers through long-term contracts, locking teams into expensive, aging hardware commitments. CAST AI is building alternatives—like tools to find spare GPU capacity anywhere—so teams don’t get stuck with “yesterday’s compute at tomorrow’s prices.”
“It’s like buying a car for three years and being told you can’t trade it in, no matter what comes next.”
Leon’s approach to leadership goes beyond the tech stack. As a brown belt in Brazilian jiu-jitsu, he draws a striking parallel between surviving on the mat and thriving in engineering teams.
“Hard things create clarity. You don’t need more meetings—you need calm minds and flow.”
His focus is on building culture through trust, repetition, and resilience. And, as he says, getting to the answer faster—without fear.
Listen to the full episode
Subscribe for new episodes at the intersection of tech, product and growth