“Processes don't fail because they're wrong — they fail because nobody updated them when the world changed.”
Greg Kabulski, Data Engineer
- 01
How has your workflow changed since the transformation started?
I build pipelines that agents consume. A year ago, I was the bottleneck — analysts submitted requests, I wrote the transform, deployed, done. Now I define the schema contract and the quality gates, and agents build the transforms against them. My job shifted from producer to governor. It's a better use of my time and the pipelines ship faster.
- 02
Walk me through a problem you solved using AI that you couldn't have tackled the same way before.
We had a merchant data enrichment backlog that was eighteen months old — tens of thousands of merchant profiles with missing category and geographic attributes. I built an agentic enrichment pipeline that classified each record using LLM inference over the merchant's name, description, and zip code. It cleared the backlog in four days. Manual effort would have been a six-month project.
- 03
What makes this different from any other company you've worked at?
The data infrastructure here is genuinely mission-critical. Every product decision, every merchant conversation, every recommendation surfaces through pipelines I own. That's unusual. Most data engineering jobs feel a step removed from outcomes. Here, I see the impact directly in dashboards within hours of a pipeline change. That feedback loop keeps the work grounded.
See open seats.
Building across engineering, product, data, sales, ops, finance, and people. Every role is an Operator role.
We get people offline through quality local experiences at great value. That's still the mission. Everything above is what it takes to deliver it in 2026.