Posted Nov 16, 2017
We seek a postdoctoral fellow for working on a joint project with Trenitalia on predictive maintenance. The position is offered for one year and is renewable for one more year.
Maintenance management in the railway environment is a key element in terms of both safety and quality of the delivered service. It has a very significant impact on the budget of railway companies. Classic maintenance is performed in the corrective or in the preventive setting. In the first case, faulty components are repaired at the first opportunity after an anomaly occurs, with negative impacts on the service. In the second case, timed or mileage-based checks are scheduled, but neither time nor mileage can accurately predict the actual wear condition of the components. In this project, we will exploit the sophisticated Trenitalia's monitoring and data collection system for predictive purposes, and we will devise and apply suitable machine learning and big data analytics tools. An accurate prediction of the operating anomalies and the wear status of the train components would allow for dynamic maintenance scheduling, with significant savings on maintenance costs, in addition to the reduction of failures and train out of service.
The AI Laboratory of the University of Florence has been active on machine learning research for three decades and has participated in several EU, national, and regional research projects. Main areas of expertise include neural networks and kernel methods, relational learning, applications to bioinformatics, neuroinformatics, natural language, medicine, and document engineering.
Trenitalia is the largest rail passenger transport company in Italy and one of the first operators in Europe, with growing profits and investments. It manages a traffic of 40 million passenger/kilometers per year over a network of over 16,700 kilometers.
The fellowship is funded by Regione Toscana and Trenitalia and comes with a full contracted positions. It offers the unique opportunity to design state-of-the-art predictive methodologies capable of scaling to very large and industrially important datasets. It also offers the opportunity to begin a long-term research career with Trenitalia.
Applicants should be highly motivated and capable of carrying out an independent research project. They should possess a good scientific publication record, hold a PhD degree in Computer Science or related disciplines, and have a solid practical and theoretical knowledge of machine learning/big data analytics and their applications. Candidates who expect to receive their PhD degree at the beginning of 2018 are welcome to apply and should send a copy of their PhD thesis together with positive evaluation reports by external referees. Candidates with excellent programming skills, preferably in Python, as well as a solid knowledge of machine learning frameworks (such as sklearn, Tensorflow, Chainer, PyTorch) will be given preferential consideration.
The fellow will be hired under a full-time fixed-term contract (assegno di ricerca), with a gross salary of 27k Euro per year. The position will start on February 2018, is offered for one year, and is renewable for one additional year. The workplace will be at the University of Florence. The fellow will be supervised by Prof. Paolo Frasconi and there will be frequent interactions with researchers in the Engineering Department of Trenitalia.
Prospective fellows are encouraged to contact Prof. Paolo Frasconi as soon as possible. A formal call will be published in December and the position will be filled as soon as a suitable candidate is identified.