Tyler
Burkett

Most people who end up at Andreessen Horowitz arrive there as pattern-recognized founders or career investors. Tyler Burkett took a different route: he built data teams at three companies across three completely different industries, watched one of them get acquired for hundreds of millions of dollars, and showed up to Sand Hill Road with actual scar tissue.

The scar tissue matters. When a16z portfolio company founder walks into a conversation about their data stack and says "we have a data problem," Burkett knows what that problem actually looks like from the inside. He's been the person staring at a warehouse that can't tell you why revenue dropped 12% last Tuesday. He's been the one rebuilding a recommendation engine from scratch when the previous one was optimizing for the wrong thing. He's been the analyst who stayed late trying to reconcile two dashboards that should agree but don't.

That was his world for most of his thirties. Before the a16z business card, there was a decade of getting his hands dirty across financial services, consumer fintech, and construction technology. Each stop taught him something different about what it takes to make a data team actually move a business forward.

Capital One: Where Analysts Learn to Think

Before analytics was cool - before Tableau decks replaced PowerPoint and before "data-driven" became something every pitch deck claimed - Tyler Burkett was learning how to think about numbers at Capital One. He spent four years there, working his way from Senior Business Analyst to Business Manager in Corporate Strategy. Capital One has always run analytics like a competitive sport; the company practically invented the idea of testing everything, measuring everything, and letting the data override the gut instinct of the highest-paid person in the room.

It's a strange place to start a career if you want to eventually help startups - Capital One is enormous, regulated, and slow to move. But it's an excellent place to learn how to structure a problem. The rigour of building credit models and strategy decks in that environment tends to stick. When Burkett eventually made his way into startups, he brought the discipline with him.

Credit Karma: Where Recommendations Actually Matter

Credit Karma was a different animal. By the time Burkett joined as Head of Recommendation Analytics in 2016, the company was already massive by startup standards - tens of millions of users, a product that monetized through financial product recommendations, and a data infrastructure that had to work at genuine scale. The recommendation engine wasn't an academic exercise. If it got the wrong offer in front of the wrong user, Credit Karma lost money. If it got the right offer in front of the right user at the right moment, that was the business.

This is harder than it sounds. Financial product recommendations have to navigate user credit profiles, product availability, lender relationships, regulatory constraints, and user trust - all while optimizing for a metric that's downstream of a dozen moving parts. Burkett led that work. Two years later, he took the skills to construction technology - which turned out to be a more interesting proving ground than anyone might have expected.

BuildingConnected: Data Inside the Acquisition Machine

BuildingConnected made software for general contractors and subcontractors to manage bid invitations and preconstruction workflows. It was, in 2018, an aggressively unsexy category to work in. Construction technology was still a frontier: a massive, slow-moving industry that barely used software, suddenly getting disrupted by a crop of startups that believed the industry's information asymmetries could be solved with better tools.

Burkett joined as Director and Head of Data & Analytics in 2018. Less than two years later, Autodesk acquired BuildingConnected for approximately $275 million. He stayed through the integration - which is the part of acquisitions that nobody talks about enough. Integrating a startup's data infrastructure into an enterprise platform is a different kind of hard than building from scratch. The constraints are different. The stakeholders multiply. The metrics that mattered at the startup don't necessarily map to the metrics that the enterprise cares about.

After Autodesk, there was Settle - a fintech startup in the accounts payable and working capital space. Another data problem. Another industry. Another set of stakeholders who needed the numbers to actually tell them something useful. By the time Burkett stepped into venture, he had handled data across four distinct sectors: financial services, consumer fintech, construction tech, and B2B fintech. That's not a typical resume for someone landing a partner role at one of the world's most powerful VC firms. It's a better one.

The title "Partner, Data & Analytics + AI" at a16z doesn't map cleanly onto what most people think of when they hear the word "partner" at a venture firm. Burkett isn't primarily a check-writer. He's the person who flies in when a portfolio company's CTO has a data problem they can't solve, or when a founder is trying to build an internal AI capability and doesn't know where to start.

a16z has always differentiated itself on the theory that the best thing a VC firm can do for its portfolio companies is give them access to expertise they can't hire for - not just capital. The model has a talent x opportunity and operating partner model built into it. Burkett is part of that operating layer. He brings the pattern recognition that comes from having built data teams multiple times, across multiple stages of company growth, in multiple sectors.

The AI dimension of the role is increasingly central. As a16z portfolio companies try to figure out what it means to be "AI-first" - not just in their product, but in their internal operations - questions about data infrastructure, data quality, and how to build feedback loops between AI systems and human decision-making are everywhere. Burkett sits in the middle of those conversations.

The Modern Data Stack Moment

Burkett arrived at a16z at an interesting time in data infrastructure history. The "modern data stack" - the ecosystem of tools like dbt, Fivetran, Snowflake, Databricks, and a growing list of orchestration, transformation, and visualization layers - was maturing rapidly. The question for every serious data team was no longer "which database should we use" but "how do we build a data culture that actually compounds?"

The technology stack listed for a16z includes Databricks, dbt, Delta Lake, SQL, Python, Hex, Fivetran, and Airtable - a set of tools that reads like a who's-who of the modern data ecosystem. Burkett works inside all of these conversations, both as a practitioner within the firm and as an advisor to portfolio companies figuring out their own data architecture.

The AI overlay is what makes 2024 and 2025 particularly interesting. Every company is trying to figure out how to plug large language models into their data workflows. The questions are genuinely hard: how do you evaluate AI outputs when the ground truth is ambiguous? How do you build feedback loops? How do you know when the model is wrong in ways that matter? These are data quality and data governance questions dressed up in AI clothing - and they're exactly the kind of questions that benefit from operator experience.

Tyler Burkett isn't a social media power user. His Twitter handle @tburk74 exists, but his feed doesn't read like a tech influencer. He posts when he has something to say. He celebrated Settle making CB Insights' Top Fintech list. He announced open roles at a16z. The posts are specific, not promotional.

That's consistent with what his career trajectory suggests: he's someone who does the work rather than narrating it. The data work that matters most - the kind that actually improves decisions at a company - is usually invisible when it's working. You notice it when it breaks. Burkett seems to understand this. He's built a career around making data systems that don't need to be noticed.

His LinkedIn handle is "its-tyler" - which is either a deliberate choice to stand out from the many Tyler Burketts who exist in the world, or a casual acknowledgment that identity on the internet is a different kind of problem than identity in data warehouses. Possibly both.

At a16z, that kind of quiet influence scales. The firm's portfolio spans hundreds of companies across every sector that technology is currently disrupting. Burkett's ability to walk into a data conversation with genuine operator credibility - to say "I've been in your position, and here's what I learned" - is the kind of resource that's genuinely hard to replicate. You can't hire for it. You have to have lived it.

He lived it across four companies, three sectors, and one acquisition. Then he brought it to one of the most powerful VC firms in the world. That's not a conventional path. But in data, conventional paths rarely produce the most interesting results.