Michael Jungmair aims to make data processing faster, more flexible and more accessible
What do you study and why are you interested in this topic?
I work at the intersection of data management, computing systems and compilers. I study how to build efficient and adaptable software systems, especially for data processing.
Data management involves the entire computing stack: from the people who want to manage and analyse data, through the algorithms and software that process it, all the way down to the code and hardware that run these tasks. My research often works across these levels to make data-processing applications faster and easier to use.
How can your research make a difference?
One goal is to make efficient data processing accessible to more people and applications. Modern data systems can be extremely efficient, but applications with domain-specific processing needs often fall outside what these systems can efficiently support. I try to close this gap so that scientists, for example, can express their ideas using familiar tools while the underlying system takes care of efficient processing.
A recent example is my work on computations that combine database operations with Python code. The challenge is to enable systems to understand and optimise such computations even when they are not written entirely in SQL.
I also develop fundamental data processing techniques and make them available through open-source implementations. For example, I lead the development of LingoDB, an open-source data processing system where we implement many of these ideas. It is exciting to see other researchers already building on it.
What is particularly exciting in your field right now?
Three developments stand out. First, computer hardware is changing rapidly, driven by physical limits and the expansion of AI infrastructure. Large-scale systems therefore increasingly combine different types of processors, such as CPUs and GPUs. However, making effective use of these different processors together remains a challenge, especially in applications beyond machine learning.
Second, some hardware resources are becoming scarcer and more expensive. This increases the importance of software efficiency and makes compilers that enable zero-cost abstractions particularly interesting.
Finally, AI is changing both systems research and how software is used. As a research tool, AI can help small academic teams prototype and explore ideas with less engineering effort. At the same time, AI agents are becoming users of software systems themselves. This raises new questions about what these users need from a system and how we can support their interactions efficiently.
What are your plans for the future?
At Aalto, I want to build a research group working across data management, computing systems and compilers. My long-term goal is to improve the efficiency of large-scale data and software systems running real applications on modern hardware.
One concrete direction I want to pursue is end-to-end optimisation of data-intensive applications in scientific computing and data analysis. I want to understand how their components interact and how improvements across the stack can make the whole application more efficient.
Read more news
Björn Ottersten appointed to Aalto University Foundation Board
Hanna Sievinen continuing on board for second three-year term
Yvonne Nygård appointed Adjunct Professor at the Department of Bioproducts and Biosystems
DSc (Tech) Yvonne Nygård has been appointed to a five-year fixed-term position as Adjunct Professor at the Department of Bioproducts and Biosystems at Aalto University School of Chemical Engineering. Her field of professorship is industrial biotechnology. Nygård will contribute to teaching.
Recruitment to begin for position of Aalto University President
Aalto University is more than a workplace. It is an open community, building a sustainable future for the coming generations.