Hi, I’m Constanza Liévanos.

Over the last few months I've been one of the core engineers on Link Planner, Taara's web app for planning and analyzing microwave wireless link availability and performance. I came in during the alpha and helped take it from a rough internal demo to something customers actually use.

Early on, my job was making the core experience real. I rebuilt the UI from scratch: the app shell, the map and link analysis views, and the inputs people use to define a link. I also turned the link itself into a real object you can create, name, edit, save, share, and export. That meant designing the whole save and share flow, the email people get when a link is shared with them, and the guardrails so nobody loses work halfway through.

From there I led the frontend for Weather Twin, one of the product's signature features. It helps people reason about a link's availability by looking at locations with similar weather. I took it from first integration through a big redesign, shifting the UI from showing every strong match to recommending a single best twin, and ran it as a real experiment. I owned both the ranking logic behind it and the analytics that let us measure whether it actually worked.

I also owned authentication for the product, which turned into one of the biggest pieces of my work this year. I built it end to end on Better Auth: the sign in and sign up screens, the backend integration, and a flow where new users can create their own account and organization. On top of the happy path I handled everything auth quietly demands. Invites, email verification, terms and privacy gating, blocking personal email domains, token validation, and a fair bit of security hardening, all backed by its own test suite.

I led the client side of a migration to a new availability API, including surfacing annual confidence ranges in the UI as stacked charts. Alongside the feature work, I kept pushing on the details that decide whether a product feels finished: refactoring toward shared, tested utilities, deleting legacy code as features matured, instrumenting the product so we could learn from it, and the long tail of small UX fixes that turn a prototype into something people trust.