I build the layer between a company's data and its decisions.
I'm currently an engineering manager at Netflix Ads Data and Insights. My team builds the catalog where the ads org defines its metrics (it sits on top of Netflix's shared semantic layer, which another team owns) and the internal analytics products those metrics show up in. If someone at Netflix is trying to understand how the ads business is performing, there's a good chance they're reading something we built. Before I managed anyone I spent five years at Netflix as a senior and staff data visualization engineer, and almost a decade in investigative data journalism before that, mostly at The Washington Post. I still think the hardest problem in data is getting people to agree on what the numbers mean.
What I do
We run the catalog as a product. Every ads metric defined in it has one definition that product, finance, science, and leadership read from, and we've moved most of the legacy models onto it so far. On top of the catalog we build the dashboards and exploration tools, the ones that answer "so what?" instead of just "what?", plus an API and MCP tooling so the same definitions reach code and agents, not just screens. The agents we've shipped audit and monitor. The one that lets someone ask a question in plain language and get a number we'd stand behind is what we're building toward.
I decide what gets built and why, and I'm trying to get to where the team makes those calls without me. I still make the technical calls that matter: how the catalog is modeled, what an agent should be allowed to ask, and where a definition belongs in code instead of a prompt, since a definition in a prompt drifts and one in code can be tested. What I've written here so far is mostly about that last part.
Writing & talks
See all →- I Spent Over $200 Teaching a Model What "Clean" Means 2026 · Blog
- More of What, Exactly? 2026 · Blog
- AI Engineering Lessons from a 52-Hour Hack Project 2026 · LinkedIn
- Codebreaking Creativity: Unlocking the Stories in the Data 2025 · Talk, Canva
- On Upward Mobility 2022 · The Pudding
- Exit Interviews: Aaron Williams 2021 · OpenNews Source
Community
Board member at the News Product Alliance, where I developed and taught the "Data for Product Management" curriculum. Former OpenNews board member.
If you've hit the same problems, or you think I'm wrong about some of it, I'm always down to talk. hi@acwx.net or LinkedIn.