About
I started out in marketing and taught myself to code because I kept needing tools that didn't exist. This is how that happened.
I build what I can't find. My day job is growth marketing at EquityMultiple, a real-estate fintech in New York, and somewhere along the way I started building the software my team needed — first scripts, then retrieval pipelines, then a whole memo platform.
My career started in growth marketing: paid acquisition, SEO, lifecycle, and conversion experiments. At HexaGroup I ran strategy for the agency's fintech and B2B clients. In 2019 I got research access to GPT-2 and started poking at what language models could do for search ads and content. I've been curious about them ever since.
The marketing side of that job is on the resume if you want the specifics. What I didn't expect was how much of it would become building. Memosa, a platform that reads a sponsor's Excel model as a dependency graph and drafts the investment-committee memo, took about 20 months. Along the way came MCP servers for the team's workflows and a steady habit of conversion experiments. The engineering turned out to be my favorite part of the job.
I studied International Relations and Chinese, which has almost nothing to do with any of this. Currently based in Washington, about 20 minutes from Portland.
How I tend to work
Start with the problem
Everything I've built started as a specific problem someone was living with — a memo eating analyst weeks, a feed I couldn't filter. The tech gets chosen after the problem is clear.
Make it real
I keep the things I build running, which means living with what I got wrong. Most of what I've learned showed up months after launch, not during the build.
Measure everything
I like knowing whether something actually worked. Experiments get success criteria before they run and a straight look at the results after — the failures usually produce the most useful notes.