All projects
Professional | 2024 — Present |
Live in production

Despegar Search Box

The search experience for Latin America's largest travel company

Role: Full Stack Developer

Rol: Desarrollador Full Stack

Overview

A micro-frontend service built with Preact and Node.js/Express that delivers the primary search entry point across all Despegar properties and white-label partners. Handles multi-product search flows — flights, hotels, packages, transfers, activities — with geo-autocomplete, date logic, and passenger configuration. Deployed on Kubernetes and served as dynamic bundles to dozens of domains across Latin America.

The Challenge

The search box is the single most critical UI at Despegar — it's the conversion funnel entry point for millions of users across different products, locales, and white-label partners. Every change must be A/B tested with real traffic, and any regression is immediately felt in conversion metrics.

The Approach

Every feature ships behind an A/B test parsed from query params. I contributed across the full breadth of the service: the Sofia WhatsApp AI integration (login-aware, modal flow, feature-flagged rollout), ML-powered recommendations in autocomplete, hotels tab for white-label partners, accommodation region support, transfers autocomplete fixes, Cypress E2E test coverage, and Hotjar analytics instrumentation.

The Outcome

Features shipped are live for millions of daily users across LATAM. The Sofia integration opened a new conversational search channel via WhatsApp. Recommendation improvements increased destination discovery engagement.

Key Highlights

01 Micro-frontend architecture serving dozens of LATAM domains
02 A/B testing infrastructure for every feature before rollout
03 Shipped Sofia: WhatsApp AI integration for conversational search
04 ML-powered destination recommendations via Euler + Freya services