The tower stood ten meters up in a stretch of Spain close to the Portuguese border so remote there was no cell signal to be found. Every twenty-four hours someone had to climb it, pull the memory card, and come back the next day. The reason? How much groundwater were the oak trees here actually drinking? Since the ecologists couldn’t measure it directly, they measured everything around it instead.

Over 10 years later, one of those ecologists, Leo Reyes, is Global Head of Data Science for R&D at WPP, where the measurement problem looks different, but is fundamentally the same. Identifiers are disappearing, client data will not leave client systems, and his organization has gone from one or two AI agents to nearly a thousand. He leads four hundred data scientists through it with an ecologist's instincts: think holistically, account for what you cannot see, and when the water keeps rising, stop building taller walls and go fix the forest upstream.

Leo brings:

• Why the fastest way to measure something is often to measure everything that it isn't
• What changes in AI governance when you hand teams a token budget instead of a policy
• Why accuracy is table stakes and transparency is the thing clients actually buy

In January 2020, Stephen Hahn, M.D., had been running the FDA for just a few weeks when the COVID-19 virus sequence was published. Despite the odds, in just eleven months, the first mRNA vaccine was authorized. It was the fastest development of its kind in history. And the part Hahn can't stop thinking about isn't how hard it was. It's what it proved: medicine can move that fast, even though it usually doesn't.

Now, as Chief Executive Officer of Nucleus RadioPharma, he makes a new argument for data science leaders: the hard part in medicine isn’t the science. It’s what we do with the data.

Stephen brings:

• Why he calls AI in molecule discovery “table stakes”
• What leading the largest real-world experiment in history taught him
• How the FDA compressed years into months, and how we can do it again

Justin Johnson sat down one afternoon to build a new capability for ARIA, the AI research agent he runs on a supercomputer that sits on his desk. Before writing a single line, he checked what was already there and found that his own system had designed and built the exact thing he was about to start. Three months earlier. While he was asleep.

That was the moment Justin went, in his words, from first mover to second mover on his own project, and he's still working out what that means.

Justin brings:

• How the “second mover” gap is widening
• Where most senior data science leaders really fall on the copilot-versus-harness divide
• How the old trade-off between mission and frontier technology is collapsing in biopharma

Be sure to check out Justin’s book, Builder-Leader: The AI Exoskeleton That Crosses the Gap, and his blog, Run Data Run.