Ably

Five engineering leaders who’ve stepped into big new roles this year, in their own words. I can’t wait to see what they build. Premjith Rayaroth is now SVP Engineering at Salesforce. At Guidewire, his focus was on “making the platform itself smarter,” and it’s exciting to think about that ambition at Salesforce’s scale. Francesco Nerieri leads engineering at Zendesk as SVP Engineering. Recommending a role there, he called Zendesk a “rare mix of incredible scope and impressive team and leaders.” That combination is where great engineering happens. Albert Orriols Puig is building Indeed's AI platform as VP of AI Platforms. His invitation to the people joining him: “build AI that actually matters at a scale that counts.” That’s a team worth watching. Kris Rasmussen moved from CTO to Chief Architect at Figma because, in his words, he’s been “itching to get closer to the work.” That’s the kind of move engineers love to see. Uday Ruddarraju leads compute engineering at OpenAI as CTO of Compute. He calls it “one of the hardest engineering and infrastructure challenges the world takes on.” Difficult problems in very good hands. Five ambitious remits, five people who clearly love the work. I’ll be following every one of them.

Here are 15 engineering leaders I think are worth following across AI, infrastructure and engineering at scale: → Katelyn Lesse leads platform engineering at Anthropic. Her work on Claude’s developer platform gets into a question more teams will face: how do you run agents within real security boundaries? → Boris Cherny leads Claude Code. I like his way of talking about coding agents as a way to shorten the distance between having an idea and seeing it work. → Ashwin Sreenivas is building AI agents for customer service at Decagon. He’s shared what it takes for those agents to look up information, take action and handle real customer requests. → Johann Schleier-Smith brings deep database and systems experience to Temporal’s work on reliable AI workflows. → Nic Cavigliano is a Technical Fellow at Indeed. I recently asked him what happens when an agent experience faces 10× growth. His answer on clear boundaries, backpressure and graceful degradation was worth sharing. → Michele Catasta writes about the engineering behind Replit Agent, including harness design and how to evaluate and improve agents at scale. → Adrian McDermott is CTO at Zendesk. He’s written about how AI is changing customer expectations and the technology needed to meet them. → Ryan S. leads infrastructure and core technology at Fin. His Rails World talk opened up the work behind keeping tests, deployments and shipping fast as agents produce more code. → Dharmesh Shah is building Agent.ai alongside his work at HubSpot. He shares his experiments with making agents useful and accessible to more builders. → Samar Abbas has worked on durable execution systems at AWS, Microsoft, Uber and Temporal. His recent writing tackles what AI agents need to keep working when a step fails. → Théo Carrive leads engineering at Mistral AI. He’s also shared practical accounts of how engineering teams work and what they learn from production incidents. → Gergely Nemeth leads Client Foundations at DoorDash. His writing on AI tools asks a good question: how do engineers use more automation without losing their understanding of the systems they build? → Fabian Hedin is co-founder and CTO at Lovable. He’s explained decisions behind the apps it generates, including the move to server rendering with TanStack Start. → Farhan Thawar leads engineering at Shopify. His perspective on AI in engineering goes beyond tool adoption to what engineers still need to understand for themselves. → Yoav Abrahami has written detailed guides to Wix’s architecture, from designing around different service levels to making microservices more reliable. Whose work would you add to this list?

"We'll build it once, then get back to the product." It's never just once. Once to handle more connections. Once to chase down delivery delays under load. Once to stop a brief disconnect from becoming a reconnect storm. Each fix makes sense on its own. Together, they turn into a permanent job for a team that thought it was done. Scoping it as a one-off project keeps the upfront cost low. The rest gets paid later, one fix at a time. "Built" is a milestone, not the finish line.

A slow failover shows up as a stalled chat, a dropped livestream, a lost trade. Ably's infrastructure is built so failure resolves before your users even notice. That's the difference between a platform that's reliable on paper and one that's reliable in production.

Stop spending engineering time on WebSocket management and homegrown pub/sub. Ably lets you offload the complexity of global messaging, fan‑out, and presence to a platform proven at handling billions of connections every month. With infrastructure built to isolate and recover from failure automatically, your team ships features instead of managing edge cases and outages.

Building realtime infrastructure in-house meant HubSpot dedicating 20% of its engineering team to it -- time better spent on core product. With Ably, they cut that overhead, saved $300k a year, and now reliably deliver 500,000+ business-critical conversations every month for 268k+ customers.