Internship: Automated tuning of airline cloud application with reinforcement learning

Description de l'offre

Join us and shape the future of travel

Shaping the future of travel has always been important to us at Amadeus. Today, with technology getting smarter by the minute, that future is more exciting than ever.

We work at the heart of the global travel industry. Amadeus offers you the opportunity to learn and grow and drive your own progression in an exciting and multicultural environment.

Our people are driven by a passion for 'Where next?' If you want to shape your career and the future of travel, Amadeus is the place for you.

Team description:

In the recent times, the complexity of Revenue Management techniques has advanced, and the large airlines with significant connection traffic. To accurately propagate the advanced revenue management logic across all the distribution channels, airlines now require the availability to be evaluated in real-time. At the same time, with the rise of Internet, the transaction volumes are increasing exponentially.

In regards to this major industry trend, the purpose of Amadeus Airline Cloud applications is to propose a new approach to distribute accurate availability and shopping in a scalable, reliable and cost efficient manner. The R-Box project aims to build a product called Airline Cloud Availability which will enable Altéa hosted airlines to quickly and massively distribute their availability, closer to where it is needed, by leveraging private and public cloud hosting. The S-Box project leverages and extends Cloud Availability, adding in support for connection build and shopping services.

The Altea Inventory Performance team, part of the Cloud Availability Platform (CAP) department, is responsible of bringing technical expertise around performance and cloud technologies (Amadeus Cloud Services), and to participate to the development, operations and optimisations of all those new products.

If you are interested in strategic and innovative projects, apply to this internship!

Main responsibilities:

As part of the Inventory performance group, you will use your software design, analysis and programming skills to help the team tocreate innovative self-tuning systems.

The team you will be part of handles a broad range of challenges from cloud deployment improvement, flexible monitoring, to both low-level C++ software and high-level functional algorithmic optimisations.

More specifically, you will have the chance to participate indesigning and implementing a reinforcement learning-based solution to improve operations of our cloud applications, and its overall performance.

We use openshift to implement availability service to airlines through docker containers.

Until now, we optimized hardware (CPU/disk) usage through slow and iterative manual fine-tuning of docker topology.

Goal of this internship is to implement classic RL components in this scope. Environment is overall system, and performance is evaluated via task reward, implemented using system response time and stability. Q-learning algorithm can be used to evaluate discrete actions to trigger. Discrete actions could be composed of openshift pods rescaling, pod blueprint selection (to increment/decrement number of service instances within pod), and even possibly changing CPU requests assigned to each container.

You will gain knowledge on state-of-the-art techs: PaaS (docker/kubernetes/openshift) applications, learning rewarding fetched from highly scalable NoSQL frameworks (Kafka/Spark/InfluxDB).

You will be immersed in agile managed teams (scrum).

Expect many challenges, including deployment on a variety of data centers (Amadeus, Google or Amazon), leveraging machine learning techniques to find optimal configurations, and significantly decreasing operation cost.

This internship will take place in new innovative Amadeus Villeneuve-Loubet campus, in modern and dynamic work environment.

Any duplication and display of partial or full content of our job advertisement on any support, such as brochures, websites, mail, emails, this list is not exhaustive, is strictly forbidden without prior formal Amadeus’ authorisation.

Recruitment agencies:Amadeus does not accept agency resumes. Please do not forward resumes to our jobs alias, Amadeus employees or any other company location. Amadeus is not responsible for any fees related to unsolicited resumes.