PhD Position in Decentralized Resource-Constrained Machine Learning
The Distributed Computing (DISCO) Group at ETH Zürich, led by Prof. Dr. Roger Wattenhofer, is seeking a highly motivated PhD candidate to join the SNSF Ambizione 2023 project “eDIAMOND: Efficient Distributed Intelligent Applications in Mobile-Network Dynamics.” The group is renowned for its research in machine learning, distributed systems, and network theory, with ongoing projects in graph neural networks, natural language processing, algorithmic learning, blockchains, consensus, cryptocurrencies, decentralized finance, and more. The eDIAMOND project focuses on developing new methods and systems for decentralized and distributed data-driven approaches to Federated Learning on resource-constrained networks. The successful candidate will design, develop, and evaluate data-driven methods, algorithms, and systems for three main research directions: distributing model training and inference over networks of resource-constrained devices; online, context-aware adaptation of federated neural network architectures based on available system resources; and communication-efficient knowledge exchange among networked federated large models. The position offers the opportunity to build an independent research profile while contributing to the project’s goals. Candidates will be responsible for the full scientific workflow, including problem motivation, literature review, system design, experimentation, and publication of results. The group values both theoretical and practical research, with regular meetings and feedback to support the candidate’s progress. Applicants should have a strong interest in the project’s topics, with theoretical and practical experience in areas such as deep learning, federated learning, distributed systems, online algorithms, neural architecture search, computer networking, and mathematical optimization. A Master’s degree in Computer Science, Engineering, Mathematics, or a related field is required, along with strong programming skills (Python, PyTorch), scientific writing experience, and excellent English communication and collaboration abilities. Additional assets include published articles, experience with network simulators, high performance computing, and Git. ETH Zürich offers a vibrant, inclusive, and diverse research environment, with numerous benefits such as public transport passes, sports facilities, childcare, and pension plans. The university is committed to sustainability and equality of opportunity. Applications must be submitted online via the ETH Job Portal and should include a motivation letter, CV, transcripts, referee contact details, and any additional relevant documents. For questions about the position, contact Dr. Antonio Di Maio at [email protected]. ETH Zürich is a world-leading university in science and technology, providing an inspiring environment for over 30,000 people from more than 120 countries.
