The man with a database in his back pocket
Most founders pitch a vision. Michael Driscoll bought one back. In 2017 Snap acquired Metamarkets, the analytics company he co-founded and ran as CTO. In 2020 he and his co-founder did something unusual: they purchased the technology back from Snap and used it as the foundation of a new company, Rill Data. The thing he had built, lost to an acquisition, and then reclaimed - that is the spine of the second act he is living now.
Rill is where his attention sits today. The premise is contrarian and a little rude: analysts are not dashboard designers, yet every BI tool on the market forces them to become one. Rill's answer is to stop asking. "Rill has an extremely opinionated view," Driscoll has said. "We just say 'hey, tell us which metrics you want to surface to your business stakeholders,' and then we auto generate the dashboards." Define the metric once. The interface assembles itself. The dashboard, in his telling, was never the point - it was the tax you paid to see a number.
That belief has a name inside the company: metrics-first. Driscoll argues that metrics, not database tables and not dashboards, are the proper interface between data teams and the people who actually run the business. His line for it is almost biblical: "Consistently defined, frequently updated, easily accessible metrics are manna for knowledge workers." It sounds like marketing until you remember he has been building the plumbing under that sentence for thirty years.
Before the dashboards, the genome
The strangest line on his resume is the first one. Driscoll began his career as a software engineer on the Human Genome Project, writing code to wrangle the largest dataset humanity had assembled at the time. He went deep into the science - a PhD in Bioinformatics from Boston University, after an AB from Harvard. Bioinformatics is, at bottom, the problem of finding signal in absurd quantities of noisy data. He has been solving versions of that same problem ever since, just with the nouns swapped out: base pairs became ad impressions became business metrics.
The detour through biology matters because it explains his instincts. People who come to analytics from genomics do not think of data as rows in a spreadsheet. They think of it as a measurement problem - how do you ask a question of something too big to hold in your head, and trust the answer? That question runs straight through every company he has built.
CustomInk, Dataspora, and the consultant years
Before he was a database guy, he was a commerce guy. Driscoll co-founded CustomInk, the online retailer for custom-printed apparel, early in his career. Then came Dataspora, a data science consultancy he founded that was later acquired by Via Science. Dataspora was the laboratory where he sharpened the craft - working across online retail, life sciences, digital media, insurance, and banking, the unglamorous industries where data either pays off or it doesn't. He spent more than a decade in that trench before the company that made his name even existed.
Those years also produced a writer. On his personal site and on Medium, Driscoll has published essays on data engineering long before the discipline was fashionable. One of them, a piece arguing that data engineers were the unsung heroes of the analytics revolution, reads now like a forecast. He saw the people who plumb the pipes as the ones who would matter, years before the job title was cool.
Metamarkets, and the accidental database
In 2010 Driscoll co-founded Metamarkets, an analytics company for the programmatic advertising world, and took the CTO seat. The early days were lean - a shoebox office off South Park in San Francisco. The company had a problem that would not go away: it needed to run fast, interactive queries over a firehose of streaming ad data, and nothing on the market could do it. For roughly a year the team threw everything at the wall - Greenplum, InfoBright, MySQL, HBase - and watched each one buckle.
Then the company's first full-time engineer, Eric Tschetter, sat down and, over about eight weeks, wrote a new database from scratch. It went into production on April 4, 2011, retiring the HBase cluster it replaced. They called it Druid. In October 2012 Metamarkets open-sourced it. Within a few years Druid was running real-time analytics at Netflix, Lyft, Salesforce, and Pinterest, and eventually became a top-level Apache project. Driscoll didn't write the code - he is careful to credit Tschetter - but he was the founder who created the conditions for it, and he later wrote its origin story so the history wouldn't be lost.
Metamarkets was acquired by Snap in November 2017. For most founders that's the ending. For Driscoll it was an intermission.
Rill: the second swing
Rill, founded in 2020, is the company he describes as the one he was building toward all along. Its products pair an embedded, in-memory analytics engine - DuckDB powers the open-source Rill Developer, while Apache Druid powers Rill Cloud - with a deliberately opinionated, metrics-first interface. The pitch to the market is blunt: it is an alternative to the Looker-and-Snowflake stack that dominates modern BI. "Over the last two years," Driscoll said around the company's launch, "we started selling this BI stack, which is an alternative to the Looker-Snowflake stack."
In August 2022 Rill announced a $12 million seed round - True Ventures, Bloomberg Beta, Sierra Ventures, Park West Asset Management, DCVC, and a long list of data-entrepreneur angels - and shipped Rill Developer as open source. The company has been fully remote since day one. Driscoll, who is also a founding partner of the venture firm Data Collective, sits on both sides of the table: the operator who ships and the investor who writes checks into the same ecosystem.
What he keeps circling back to in interviews is the next user of all this machinery. Not the analyst. Not the executive. The agent. Driscoll has argued that AI systems are increasingly becoming the primary consumers of data tools - that the clean, consistently defined metric layer he has been preaching is exactly what a machine needs to reason about a business without hallucinating. The metrics-first gospel, it turns out, was built for readers who were never going to be human.
Thirty years in, the through-line is clear. From genome to ad exchange to boardroom dashboard, Driscoll has chased the same stubborn idea: make the number fast, make it trustworthy, and get everything else out of the way.