at Rocket Travel, Inc in Chicago , Illinois, United States
Job Description
Duties: Build machine learning models using algorithms like random forests and gradient boosted trees to automate the process of mapping hotel entities from our 40+ wholesale suppliers to our master hotel database. Leverage Amazon’s Mechanical Turk to crowdsource human workers at scale to verify machine-learned hotel mappings flagged as “low confidence”, carefully designing the study using probability and statistics to maximize return on investment (i.e. obtain as many verified mappings with satisfactory accuracy given fixed budget). Use Python to programmatically retrieve hotel static content data from our partners’ application programming interfaces (APIs) like Giata and Booking.com to use as input for our internal mapping models. Collect, store, and query against region shapefiles (i.e. data defining region boundaries) using geographic information systems (GIS) tools such as ArcGIS and CartoDB to create and maintain our “geofencing” logic (which maps hotels to regions) in top regions. Create data pipelines using Python, Spark, and Airflow to compute daily/weekly/monthly metrics on hotel pricing, availability, searches, and bookings from transactional data. Build internal dashboards to visualize and monitor various supply and booking metrics, reporting key metrics in a monthly report to leadership team and company at large. Use time series anomaly detection techniques to automatically identify sudden changes in various hotel supply and booking metrics, escalating anomalies and resolving bugs in a timely manner. Write complex ad-hoc Structured Query Language (SQL) queries in MySQL and Redshift databases to extract and analyze customer behavioral and transactional data to help product and supply teams make data-informed decisions. Programmatically analyze gigabytes of historical auction pricing data to simulate “what-if” scenarios (e.g. how many bookings would a supplier gain if it reduces its price by 1%)
