•
Own ambiguous customer or internal-team problems from discovery through measurable production outcomes
•
Build coherent vertical slices across frontend, backend, workflow state, data, testing, observability, security, and release
•
Make complicated processes clear without hiding exceptions, uncertainty, or recovery paths
•
Diagnose difficult production behavior and remove recurring root causes
•
Establish useful measurements and improve customer, team, quality, or reliability outcomes
•
Create reusable product and engineering capabilities that make later work faster and safer
•
Use AI deeply in development and where it improves the product, with explicit evaluation and verification
•
Work directly with Product, Design, customers, domain experts, and other engineers
The common thread across our product is turning difficult, failure-prone work into software people can understand and trust. These areas share one Full-Stack Product Engineer title and hiring bar, but they represent distinct work and ownership. You do not need to choose one when applying. We will discuss the most relevant current opening early, and the placement-specific portion of the interview will reflect that work.
Bring fragmented enterprise data into one trustworthy system
Customers need to bring internal work data out of many SaaS tools, APIs, files, and export processes. Build the product that takes them from “our data lives over there” to a successful, verified transfer. You might design provider-specific export journeys, model long-running transfer state, make failures and recovery understandable, or build contracts, fixtures, and acceptance tests that keep every new source from becoming a one-off. This also means maintaining clear provenance and health across every handoff.
Help companies navigate dissolution from start to finish
Build the software that helps a company wind down its operations responsibly. The product spans onboarding, forms, documents, auctions, permissions, government obligations, operational closeout, and the exceptions that appear along the way. You might model durable workflow state, generate or parse documents, reconcile conflicting information, design a safe team intervention, or make a consequential next step clear to a customer. The challenge is keeping the whole journey understandable and recoverable when reality deviates from the expected path.
Make the de-identification pipeline understandable and trustworthy
Our pipeline turns sensitive enterprise data into de-identified datasets without losing useful structure and meaning. Build tools that show what ran, surface what was missed or changed incorrectly, and support delivery decisions. You might seed synthetic data with subtle failures, replay past defects, construct golden examples, combine deterministic checks with bounded model-based judges, or design investigation interfaces that reveal what no single metric can. The goal is evidence the team can interrogate and trust.