Andy Pavlo Joins ClickHouse to Launch ClickHouse Labs

August 3, 2026 | 5 minute read

I am thrilled to announce my transition to ClickHouse, where I will be establishing and directing a brand-new research initiative: ClickHouse Labs.

📜 The Origin Story

My journey began in 2013 as a professor within the Computer Science Department at Carnegie Mellon University. Throughout my professional life, I have been driven by a desire to decode the science behind modern database management system (DBMS) internals. I make it a point to monitor every emerging system—whether from academia or industry—to dissect their specific implementations.

I first encountered ClickHouse when it debuted as open-source software in June 2016. At the time, my internal monologue was: ~~"This is definitely vaporware."~~ It simply seemed too advanced to be real.

The 2016 Landscape

ClickHouse possessed capabilities that were, at the time, exclusive to a tiny number of expensive, closed-source analytical systems.

Feature	Typical 2016 Open-Source DBMS	ClickHouse (2016)
Language	Primarily JVM-based	`C++`
Execution	Standard processing	Vectorized query execution
Optimization	Lacked SIMD support	Integrated `SIMD` optimizations

Since that revelation, I have kept a close eye on their progress. ClickHouse has remained a vanguard system, deeply relevant to our academic inquiries. In fact, during the 2020 pandemic—when my database courses shifted to a remote format—I was frequently seen wearing my original ClickHouse t-shirt during lectures.

Because of this long-standing admiration, I was deeply honored when the co-founders invited me to build this research group. Joining one of the world's most elite engineering teams to shape the future of database tech was an opportunity I couldn't ignore. To put it in musical terms: this is a high-level collaboration akin to Killer Mike and El-P forming a hip-hop supergroup.

🎯 The Mission of ClickHouse Labs

The primary objective of ClickHouse Labs is to build a premier industry research organization dedicated to databases.

"We will not operate as an isolated research organization that throws ideas over the wall to engineering."

Instead, we are building a symbiotic ecosystem:

Strategic Focus Areas

Collaborative Innovation: Working alongside engineers and partners to ensure ClickHouse remains on the bleeding edge.
PostgreSQL Integration: Partnering with the PostgreSQL team to ensure their managed service becomes the gold standard for reliability and speed.
Hybrid Research: By leveraging both PostgreSQL (transactional) and ClickHouse (analytical), we have a comprehensive foundation to solve a wide array of database problems.

Our goal is to produce research with genuine scientific value—represented by the pursuit of optimal complexity $\mathcal{O}(f(n))$ —and translate those theories into tangible technology for the end-user. I aim to mirror the impact of legendary institutions like IBM Research and Microsoft Research, which proved that corporate labs can advance fundamental computer science while simultaneously powering commercial products and mentoring new researchers.

🚀 The Path Forward

ClickHouse already boasts a stellar reputation for technical transparency. Since the company's inception in 2021, the team has released exhaustive implementation guides and the highly influential 2024 VLDB paper detailing the core architecture. These documents are so high-quality that I actually assign them as required reading to my CMU students.

However, there is a significant "idea backlog"—optimizations and concepts the engineering team has explored but hasn't had the bandwidth to fully validate for production.

Immediate Action Plan

Accelerate the validation of the existing "idea backlog."
Transition validated research into production-ready features.
Explore the intersection of DBMS and AI/Agentic technologies.

Regarding AI, we are investigating two primary vectors:

DBMS $\rightarrow$ Agents: How should a database be structured to better serve AI agents?
Agents $\rightarrow$ DBMS: How can agents automate and optimize the development of the DBMS itself?

// Conceptual goal: The evolution of the DBMS
void evolve_database() {
    while (technology_exists) {
        research_new_algorithms();
        implement_simd_optimizations();
        integrate_ai_agents();
        deploy_to_users();
    }
}

We are exploring every possibility: new hardware, novel algorithms, innovative data structures, and fresh execution strategies. While the answers aren't yet clear (which is the essence of research), I am confident that ClickHouse's commitment to a solid relational model provides the perfect springboard for these data-intensive workloads.

I have spent my career studying how these systems are built and teaching the next generation of architects. With ClickHouse Labs, we now have the platform to advance both the technology and the people.