University of ChicagoB.S. with Honors, Computer ScienceB.A., Creative Writing
GPA 3.75. Margaret C. Annan Undergraduate Award in Writing; People's Choice Award at the Tech Showcase.
I'm Sarah Katz, a full-stack engineer in Queens, New York. I'm looking for a role with high ownership, where I solve real problems for users and build products that matter. I'm open to roles as a software engineer or technical product manager.
A board of lights in my apartment that knows when my train is coming.
It shows the next R and M trains from 63 Dr–Rego Park toward Manhattan, and once a minute the arrivals slide away and the time slides in. On Friday evenings, Saturdays and holidays it also shows candle-lighting and havdalah times. It was my first hardware project: I'd never touched a Raspberry Pi before this.
Photograph a receipt, then ask about your spending in plain English.
An AI agent reads each receipt photo into a database, line by line. Then you can ask it things like "how much did we spend on groceries in August?" and get an answer. I built it with a collaborator as a web app and I'm now turning it into a phone app that a whole household can sign into.
Describe your symptoms, and see what the visit is likely to cost you.
Patients walk into a doctor's office with symptoms but no way to know what they'll owe. With two classmates in AI Agents for Science, an elective at UChicago, I built a multi-agent pipeline that follows the whole path from symptoms to bill. A diagnostic agent turns free-text symptoms into ranked ICD-10 diagnoses. A procedure agent plans the likely procedures, finds their ICD-10-PCS codes with keyword and lexical search, and maps them to CPT/HCPCS billing codes. A billing agent then applies Medicare fee schedules and insurance terms like the deductible and coinsurance to project the patient's out-of-pocket cost.
Grounding the agents in real code catalogs mattered: retrieval-augmented mapping reached 74% top-3 accuracy, well above asking the model alone.
First engineer at an early-stage maritime monitoring startup, owning the computer vision pipeline end to end. I replaced an off-the-shelf detector with direct YOLO inference, an in-house multi-object tracker and an event manager that turns raw harbor footage into discrete vessel events. I built the service that matches camera tracks to ship transponder (AIS) data, narrowing about 444K candidate pairs to about 27K a day, and set up the company's first accuracy review. That baseline led to four measured fixes, one of which recovered 1,900+ dropped tracks.
Led REST API integrations on Google Cloud Functions for onboarding and data pipelines, handling about 100 backfill and update requests a day. Shipped an iOS demo app in Swift used in fundraising, and wrote the monthly technical newsletter for 200+ investors and waitlist users.
Full-stack Java and Ruby on Rails work on loan origination and customer-facing workflows. Wrote unit, integration and end-to-end tests with RSpec, Capybara and JUnit, and queried MariaDB/MySQL for analytics and CTO reporting.
Managed a portfolio of Chicago client organizations: outreach, scoping, and overseeing one or two technical projects a quarter from start to handoff, and mentoring project leads on stack choices.
Led development of the team's digital archive, covering 15+ years of history and records. Built the backend in Java and Flask and the front end in React, Node.js, HTML and CSS.
When COVID closed schools, I raised over R70,000 to buy 20GB of mobile data for each of 108 students so they could keep learning remotely.
My writing won the Margaret C. Annan Undergraduate Award at UChicago. I write profiles of people doing interesting work, and essays on old books, especially medieval ones like Le Roman de Silence and Chaucer. In both, I'm interested in how people make meaning out of what they're handed. Writing is also how I think through a product before I build it.
Hiring, or building something interesting? I'd love to hear about it.