What happens to human capability when AI changes how we do the work?

Learning Culture & Developer Growth

In my studies, Learning culture predicts developer productivity and team effectiveness, with every improvement on learning culture connecting to 25–30% gains in effectiveness at the team level. Software development innovation rests on developers taking advantage of the collective intelligence of your technical ecosystem; effective AI workflows depend on your organization's developer expertise setting the right context for the tools. When your environment punishes learning, developers stop sharing knowledge, stop taking creative risks, and stop growing. I call this Learning Debt, and it has compounded silently for many organizations. But you can change it.

This isn't soft stuff. Improving this is one of the most powerful levers you have. Start here.

Developer Experience That Means Something

Self-report data is phenomenal when it's designed well and garbage when it's just designed to reinforce an existing narrative. I build measurement instruments grounded in psychological science and co-designed with developers, not corporate agendas, to capture the things your developers see making a real difference — code review anxiety, AI skill threat, team affordances, belonging, psychological safety — and connect them to the outcomes your leadership actually cares about. Combined with software and business data, we show the value of the good team practices you're already trying to build, and give you and your teams more room for the human work that makes technology innovation happen.

Human beings are great reporters of their environment. Let's use that instead of pretending Jira tickets measure happiness.

Tool & Initiative Evaluation

Before you roll something out to a thousand engineers, you need to know whether it works and why. I design rigorous evaluations — intervention studies, observational causal inference, the works — that give you defensible evidence instead of a vendor's marketing deck.

Because "our numbers went up for the week after we rolled it out" is not real causal understanding.

Making Sense of Messy Software Data

Your org has more data than it knows what to do with, and less of it is clean than anyone admits. At the same time, software metrics provide a tremendous opportunity to genuinely evaluate the impact of AI adoption and practice changes. You just need the right expertise to unlock it and avoid common data pitfalls that keep most of your industry peers mired in misleading signals. I bring causal inference, multilevel modeling, and fifteen-plus years of working with longitudinal behavioral data across hundreds of thousands of learners and thousands of software developers. Big data is terrible when you don't know what's really happening. I'll help you find out.