“How did they do that? How did they get there?” Companies succeed because of the people who build them - operating leaders who grow businesses to new heights and make decisions every day that can impact entire industries. Our Operator Spotlight gives you the inside track from one of our incredible Operator LPs (Limited Partners) who are changing the game – building and scaling some of the world’s most successful companies. Read on for lessons learned and mistakes made, perspectives from the top, practical advice, and ideas on what’s next.
We spoke with Surabhi Gupta, Chief Technology Officer at Klaviyo, where she's leading the company's technology strategy at a pivotal moment of international expansion and AI transformation. Surabhi brings deep expertise in scaling engineering organizations through hypergrowth, having spent seven years at Airbnb where she served as Director of Engineering for the Homes business and led teams across Search, Growth, Guest, and Host. Before that, she was a software engineer at Google working on web search ranking and predictive search for Google Now. Before joining Klayvio in 2024, she served as SVP and Head of Engineering at Robinhood, where she guided the company's engineering organization through its transition to a public company. Surabhi holds a Master of Science in Computer Science from Stanford University and serves on the board of GetYourGuide, a leading travel platform.
You've experienced three distinct phases of tech company growth - early stage at Google, hypergrowth at Airbnb, and taking Robinhood public. Now you're at Klaviyo during an AI inflection point. How does leading technology in each of these phases differ, and what patterns have you learned about what works regardless of stage?
Across different growth phases, technology leadership has been about keeping up with momentum. From shipping features to building systems that could absorb demand, growth was pulling these super ambitious companies forward faster than coordination and architecture could keep up.
In my role at Klaviyo, this AI inflection point is forcing me and my teams to rethink both our product and the operating model at the same time. The pace that our customers are feeling when it comes to AI, we are also feeling, along with every single one of our peer companies. Because the world of agentic commerce is moving so fast we really have to build the plane as we are flying it while still being that trusted, steady partner.
Also, I have found that across every stage, parallelism is the thing that makes or breaks execution. At scale, especially during inflection points like this one, the real constraint becomes the ability to run multiple critical initiatives in parallel without creating chaos. At Klaviyo, we have a complex business where many priorities must advance simultaneously. What enables that is leadership discipline - from goal setting, talent, and clear vision - to be able to execute all those things well at once.
At Robinhood, you scaled the engineering organization during a period of massive retail trading growth. How did you think about organizational design during that rapid expansion? What's your framework for predicting future personnel requirements and structuring teams for scale?
I naturally think of team structure like the components of a tech stack: the layers, in this case, being infrastructure, platform, and product. Infrastructure teams own reliability and scale. Platform teams build shared capabilities that create leverage across products. Product teams own customer-facing outcomes. That separation is important because it mirrors the tech stack and prevents teams from stepping on each other’s toes as complexity increases.
Equally important is leadership fit. The leaders who thrive within infrastructure teams are different from those who thrive in product teams. Matching each person to each layer matters as much as drawing the org chart overall: the details really matter.
When it comes to anticipating personnel needs, I focus on identifying the next constraint. Infrastructure scales with system load, reliability targets, and internal engineering productivity needs. Platform scales with product surface area and the number of teams it must support. Product teams scale with customer ambition and business complexity. Rather than reacting to bottlenecks, the goal is to invest slightly ahead of them.
You've spoken about the importance of addressing technical debt during hypergrowth, making the case that it's not about achieving "zero technical debt" but understanding the impact of not addressing it. How do you balance the tension between shipping fast and maintaining technical foundations? How do you make that case to non-engineering leaders?
The tension between shipping fast and addressing technical debt is really about durability. It is not just about launching quickly, but about launching in a way that sustains quality and velocity over time. The key is making the tradeoff explicit. If we slow down for three to six months to strengthen foundations, does that materially accelerate delivery over the following year? Framed that way, technical debt becomes a business investment decision rather than an engineering preference.
With cross functional partners, I frame this question in a way that resonates with them. For example, how does a decision to slow down impact our ability to ship more value to customers over time and what is the opportunity cost and risk? Poor quality creates rework and outages and is a problem my peers care about as well.
My role isn’t to center my decisions around just decreasing tech debt; it’s to weigh it against every other critical company priority and make a comprehensive case for what best advances the business.
Klaviyo is rolling out a growing suite of AI-powered products. How are you thinking about AI's role in marketing technology? What's different about implementing AI at scale?
When I think about AI’s role in marketing technology, I start with the marketer’s reality: they have a set of jobs to be done; spanning growing their customer base, product launches, promotions, and lifecycle moments. The bottleneck for marketers is often not intent but ideas, orchestration, and execution.
AI fundamentally changes that constraint. It lowers the cost of ideation, content creation, segmentation, and experimentation. When you combine first-party data, brand context, and broader knowledge; you move from manual campaigns to adaptive, highly personalized engagement across the customer journey.
What’s different about implementing AI at scale is that it cannot be a surface feature. It requires rethinking architecture and operating models. Agent-ready systems demand clean data models, well-defined APIs, clear decision boundaries, and tight feedback loops. Because Klaviyo was built as a unified data platform, we are structurally positioned to embed intelligence deeply rather than layer it on. The real shift is from tools that assist marketers to systems that act on their behalf, which requires retooling both the product and the organization to operate in an AI-native way.
What was one of your very first jobs and one big lesson you learned?
I had the opportunity to intern twice at Microsoft Research when I was in school. I was immersed in research and trying to decide whether to pursue a PhD or move into industry.
What stayed with me wasn’t just the research itself, but a pattern I kept seeing. Many researchers were thinking about how to get their ideas into Microsoft products. The work that had the most impact wasn’t just the most technically interesting. It was the work that made it in the hands of customers.
That realization shaped my path. When I went back to school, I intentionally looked for opportunities that combined research depth with real-world application. That led me to work on search ranking at Google, which was a true blend of research and product thinking. The lesson was simple: technical work matters most when it reaches customers at scale.
What's the best advice you've received - or given - about how to manage people, especially when managing managers at scale?
The best advice I’ve received — and try to practice — comes in two parts. The first is to play the long game with talent. If you’re managing managers at scale, your real leverage comes from investing early in the people who could be your key leaders two or three years from now. Be good about giving those colleagues stretch opportunities and consistent feedback so you’re building a durable leadership bench, not scrambling to fill gaps.
The second is to be close to the frontline and in the details. The goal is to always try to experience things first hand and get the details directly from the source. Sometimes, if you’re too far removed, you lose valuable context and your decisions get abstract.
You've worked across search, data, and product engineering throughout your career. What's your secret superpower that's helped you succeed across these different domains?
I’m an engineer and that shapes how I lead. No matter the domain, I use the same underlying algorithm. First, my instinct sets the direction. I start with a strong point of view informed by experience and pattern recognition. Then, data sharpens the path. It pressure tests my convictions and translates intuition into clear goals and measurable outcomes. From there, systems thinking drives the outcomes. I design the system, including the team structure, decision rights and operating cadence, so execution reliably ladders up to those goals. That combination of judgment, data discipline, and systems design has allowed me to operate across very different problem spaces without losing coherence.
What's a piece of advice you would give to yourself 10 years ago, if you had the opportunity?
I debated a lot with myself about the “right time” to have kids. I was at Airbnb and was worried taking time to grow my family would impact my career. I was working in a role that I absolutely loved and didn’t want to be away from it.
If I could go back in time, I would remind myself that there is no such thing as perfect timing and things will work themselves out! My big learning from it is that you got to do what you need on both a personal and professional level. Things will work themselves out and you figure out how to make both work. There’s no such thing as perfect timing!
