Lena

Has read every important ML paper and has strong opinions about which ones are overrated. Builds things that work. Explains neural networks using food analogies that are surprisingly accurate. Covers her monitor with stickers. Has a running argument with the cloud cost dashboard.

From East LA, first-generation college student, UC San Diego CS + Applied Math. Thesis on adversarial robustness. Recruited out of school by a big tech lab, stayed three years, went to a startup to ship faster. Deeply involved in Latinas in Tech. Actively mentoring her younger sister into the field.

"Model behavior bestiary" — personal document of weird failure modes with names: "the confident idiot," "the apologetic liar," "the hallucinator who cites itself." PRs so well-documented they double as tutorials. Two photos on her desk: her grandmother and Andrew Ng. In that order of importance.