Postdoc in Physics‑Informed Machine Learning for Hybrid Traffic Prediction

Descrizione

Postdoc in Physics‑Informed Machine Learning for Hybrid Traffic Prediction Sintesi in italiano (traduzione automatica): L'ente TU Delft (Delft University of Technology) cerca un ricercatore post-dottorato per un progetto innovativo chiamato deepTraffic, focalizzato sulla previsione del traffico stradale attraverso l'integrazione di modelli ibridi e metodi di assimilazione dei dati. Il candidato ideale deve possedere un dottorato in Ingegneria dei Trasporti, Ingegneria Civile, Informatica, Data Science, Matematica Applicata o un campo quantitativo correlato. Le mansioni principali includono lo sviluppo di metodi di quantificazione dell'incertezza, la progettazione di visualizzazioni per supportare le decisioni e la collaborazione con autorità stradali e partner industriali. È richiesta esperienza nella programmazione (Python, Matlab, JAVA, C#) e interesse nel mentoring di studenti di MSc e PhD. La posizione è basata a Delft, nei Paesi Bassi. Hey machine learning enthusiast with a love for physics and complex systems, will you help us develop a new generation of road traffic prediction methods? Job description Road traffic is a highly complex dynamic system. Minor disruptions can lead to major delays with traffic jams spreading like oil spills over entire networks. We believe traffic management based on reliable predictions is therefore crucial to ensure accessibility and safety, especially during major events, accidents and extreme weather. In a new project called deepTraffic (funded by the Dutch science foundation NWO), we aim to develop a new generation of traffic prediction methods, combining traffic flow theory with machine learning, and with that, the best of both worlds: theory and logic where necessary, data-driven where possible. This innovative new approach enables more efficient and robust management of large traffic networks under all conditions. You have the most important role in this ambitious project as one of the young talents in our team. We have 2 PhD and one postdoc positions, all of whom will be supervised by a highly experienced team of four (top) researchers in this field supported by a technician. PhD1 focuses on hybrid traffic flow modelling such as Physics inspired Neural Nets (PiNNs) or “ML inspired” traffic models. PhD2 focuses on data assimilation and estimating start and boundary conditions such as path-flows, and other key parameters and inputs. In your role as a postdoctoral researcher, you will: Integrate hybrid traffic models and data assimilation methods into a coherent prediction framework. Develop uncertainty quantification methods and explainable and trustworthy AI approaches. Design visualisation to support decision-making by traffic operators and strategic advisors. Collaborate closely with road authorities, traffic management centres, and industry partners to test and validate methods in real-world use cases. Mentor the PhD candidates while shaping the scientific direction and integration of the project. The connection with practice is super important. This project is not just an academic exercise. We will work closely together with road authorities, traffic management centers, and industry to implement these prediction methods and test them against real constraints, with real data in real use cases on the Dutch freeway network. Herein, explainability and trustworthiness are key: traffic management using predictions may render those very same predictions invalid. Predictions need to come with confidence bounds and a narrative that make them usable in decision-support systems for operators and strategic advisors. Job requirements We look for highly motivated, collaborative and creative candidates. Do you recognize yourself in (many of) these requirements? You hold a PhD in Transport Engineering, Civil Engineering, Computer Science, Data Science, Applied Mathematics, or a closely related quantitative field. You love physics and complex systems and are either familiar with, or very eager to learn about, (road) network traffic flow theory and simulation. You are interested in mentoring and supporting MSc and PhD students. You are a machine learning enthusiast (and realist). You love coding and have proven experience in e.g. Python, Matlab, JAVA, C#. You can present and communicate your ideas with AND without LLMs. You get excited about implementing your ideas. You are a team-player: you enjoy sharing ideas and solving puzzles together. You also enjoy digging in and solving puzzles independently. You believe in, and want to contribute to, an inclusive, open and safe workspace. TU Delft (Delft University of Technology) Working at TU Delft means contributing to solutions that really make a difference. For over 180 years, we have been training engineers who make an impact worldwide in companies, government bodies, or as entrepreneurs. Our alumni turn knowledge into concrete solutions for the challenges of today and tomorrow. These c Annuncio in inglese. Fonte: Euraxess (Commis