Team Description:

Ever wonder why the price of a flight seems to change every time you look? The answer lies in field of airline revenue management. The Amadeus revenue management research and development team designs and implements the systems airlines use to make decisions about the prices of their flights. Our airline customers use our solutions to manage hundreds of millions of euros in revenue across thousands of flight departures per day.

Project Description:

Traditional Revenue Management (RM) systems require forecasting the market demand from past observations. The classical approach to solve this problem consist in fitting a parametric model describing the demand by finding the set of parameters making the historical booking the most “likely”. However, what happens when no or few observations are available for specific markets? One way to proceed is to “sponsor” the new market with a “donor” forecast’s estimated parameters. The two main tasks in the process of sponsoring are:

• Selection of the donor itself

• How to use the donor estimated parameters accurately to ensure alignment between donor and sponsor’s forecast demand.

However, some issues as mismatch between donor and sponsor demands can be observed because of difference between features values like yields.

Several studies involving analytical formulas to recalibrate feature values, and therefore, reduce the mismatch in forecast demand are currently underway. The idea would be to investigate in GenAi and LLMs to see if there could be opportunities to improve the sponsorship process.

The internship will involve:

• Familiarization with the basic concepts of airline revenue management.
• Review of the state of the art of GenAI/LLM applied to Revenue Management.
• Familiarization of existing experiments developed internally.
• Explore features embedding methods and, more generally, new ways of matching routes (Origin-Departure). To be compared to today’s production method.
• Investigate into other kind of features.
• Feature engineering using LLMs to improve the donor/sponsor mapping and reduce the eventual mismatch in distribution that could be observed.

• Design new KPIs to evaluate accuracy of proposed and existing methods.

The ideal candidate will possess:

• Strong communication skills in English.
• Demonstrated aptitude and interest in statistics and mathematical modelling.
• Familiarity with at least one Machine Learning / Deep Learning library.

• An interest in the airline industry is a plus.

Diversity & Inclusion

We are an Equal Opportunity Employer and seek to hire the best candidate regardless of age, beliefs, disability, ethnicity, gender or sexual orientation.