Research and development in optimizing data systems for multimodal AI workloads including SQL-queries and machine learning algorithms. Be a major driver of the next-generation data systems.

School: School of Engineering
Starting date: 1.1.2027

Recent advances in AI are driving applications that integrate large language models (LLMs), vision models, vector search, retrieval systems, and large-scale analytics. These multimodal workloads run across heterogeneous platforms, including distributed processing frameworks, vector databases, ML runtimes, and GPU-based systems.

In collaboration with IBM Research Europe – Zurich, we investigate next-generation data systems for multimodal AI. The project explores architectures that dynamically orchestrate tasks across heterogeneous engines, optimize execution, and manage resources across CPUs, GPUs, and specialized accelerators.

This postdoctoral position focuses on optimizing and executing multimodal AI workflows involving LLMs, vision models, and data analytics across multiple processing engines.

Your work may include:

• Developing architectures that support multimodal AI pipelines.
• Investigating optimization techniques for AI pipelines, including analytical- and learning-based optimizers for heterogeneous systems such as relational databases (e.g. PostgreSQL, DuckDB) and distributed data processing frameworks (e.g. Spark, Presto).
• Building resource management and scheduling mechanisms for heterogeneous hardware environments (CPUs, GPUs, hardware accelerators).
• Enabling efficient execution of AI workflows, where multiple models and tools interact dynamically to solve complex tasks.
• Implementing prototype systems and evaluating them on realistic AI and data workloads.
• Assisting in teaching (Databases and Intelligent Information Systems)

The research will bridge ideas from database systems, distributed systems, and machine learning systems.

Required qualifications:

• PhD degree in Computer Science, AI, Data Engineering, or a related field
• Strong background in distributed systems (e.g. Spark), relational databases (e.g. Postgres or DuckDB) and machine learning
• Solid programming skills in Python
• Interest in publishing in leading systems and AI systems conferences such as VLDB, SIGMOD, ICDE, etc.

Preferred experience:

• Machine learning-based query optimization
• Modern agentic AI frameworks
• GPU or accelerator-aware systems
• Good communication skills in German are a strong plus (mainly required for teaching assistance)

Zurich University of Applied Sciences ZHAW is one of Switzerland's largest multidisciplinary universities of applied sciences, with over 14'000 students and 3'400 faculty and staff.

School of Engineering (SoE), as one of Switzerland's leading educational and research institutions, focuses on future-oriented topics. Fourteen institutes and centers guarantee high quality education, research, and development with a focus on energy, mobility, information, and health. The InIT Institute of Computer Science is one of the largest institutes at ZHAW and enjoys an excellent reputation in computer science and data science, both in Switzerland and internationally.

ZHAW is committed to gender-mixed and diverse teams in order to promote equality, diversity and innovation.