Meet the team

I am a founding engineer at Kumo.ai, working on bridging the gap between our predictive query language and graph learning architectures. I obtained my PhD in Computer Science from TU Dortmund University in 2022, doing research on 3D computer vision using graph neural networks. I also hold a research group leader position at the Max-Planck Institute for Informatics, heading the geometric representation learning group.

My first contact with graph neural networks was in 2018 when I worked on efficient GPU algorithms for message passing, which laid the foundation for Pytorch Geometric (PyG), mapping large-scale graph neural networks to parallel computing hardware. I worked on applying GNNs in several different areas of computer vision, graphics, and molecule analysis, making use of the versatility of the message-passing formulation. I joined Kumo in 2021 as a founding engineer to bring graph neural networks and PyG into real-world products.

It is an amazing experience to work in Kumo's international team of highly skilled and experienced people, getting to know the industry and learning from the best. In contrast to academia, creating a product for the real world holds a whole list of different challenges. It is exciting to think about and solve these problems, especially if the venture takes the team to entirely novel grounds, which regularly happens at Kumo.

Outside of working I really enjoy snowboarding and skiing if the season allows it! If it doesn't, I like watching shows, playing games or working out. You can reach out to me on LinkedIn.

I’m a founding engineer at Kumo.ai, and I’m working on the machine learning platform that makes machine learning models faster and more scalable. I got my M.S degree from UCLA and B.S from Tsinghua university. Before Kumo.ai, I was working on machine learning for self driving cars, search, and recommendation systems.

I was involved at the beginning of Kumo.ai and mainly focused on bridging the gap between the machine learning models and the infrastructure. There are unique platform challenges when we adopt GNN to real world applications coming from both algorithms and data scalability point of view. I was excited to see how I can optimize the end to end pipeline from CUDA kernels to distributed system design.

What excites me the most about working at Kumo.ai  is the mission of the company. I have a deep interest in applying machine learning to daily applications and I’m lucky to be working with a group of talented people in academia and industry to make this dream come true. Beyond daily platform development, I’ve learned a lot about the latest trends in machine learning and challenging problems in different industries. The variety of problems we are facing at Kumo.ai is unique compared to other companies.

In my free time, I love building Lego and watching movies. Connect with me on Twitter, or LinkedIn.

I am a founding engineer at Kumo.ai and work on neural architectures for the Kumo.ai platform, as well as explainable AI for Kumo.ai (and PyG). I graduated from the Department of Computer Science at Stanford University with a PhD degree, advised by Jure Leskovec. My PhD focus was on graph representation learning and graph neural networks.

I leverage GNNs in a variety of real-world applications, including recommender systems, anomaly detection, knowledge graphs, scientific simulations, molecule prediction etc. I have been a long-term user and contributor to PyG when developing all the research projects, and aim to create a GNN-based platform that can help users to achieve their goals with graph learning in an easy-to-use manner. Thus I am excited to lead many efforts in PyG for the research community, as well as development in Kumo.ai for industry users.

Working and creating Kumo.ai have been an exciting journey, where we gradually shaped the product from ideas through multiple iterations of brainstorming, prototyping and experimentation. I enjoy working in an agile environment where all of us are quick in responding to customer requests and experimenting with novel ideas in quick iteration to find the most effective and generalized solutions.

I recently joined Yale University as an assistant professor, but I am continuing as a technical advisor and contributor to Kumo.ai. I am excited to bring my new research in trustworthy GNNs, graph pre-training and AutoML to PyG and Kumo.ai. In my free time, I like hiking, piano, and PC building. Reach me at Twitter.

I’m a founding engineer at Kumo AI, and I mainly work on core graph machine learning algorithms and end-to-end AutoML solutions. I graduated from Stanford with my CS Ph.D. degree, advised by Prof. Leskovec.

I was involved in the early ideation of Kumo AI. The open-source project GraphGym that I created, is now a core component in PyG, echoing the mission of Kumo AI to democratize AI to domain experts. With GraphGym, I demonstrated that graph learning methods could perform on relational databases significantly better than traditional ML solutions. Moreover, users would be able to use the pipeline with minimal effort thanks to the ML design space and AutoML algorithm. I was excited to see that these early explorations played an integral role at Kumo. It was natural for me to join Kumo AI and make an impact in the real world with these technologies.

What excites me the most about working at Kumo AI is the team’s strength and how we collaborate. At Kumo AI, I’m lucky to be working with some of the brightest minds in academia and industry. Beyond contributing to the platform development, I’ve learned and contributed in significant ways to everything from recruiting and logo design to customer engagement and product design. This type of multi-tasking is uniquely encouraged at Kumo AI!

In my free time, I love traveling and photography. I’ve visited more than 20 national parks. Fun fact: I skipped 2 years in my primary school. Connect with me on Twitter, LinkedIn, or Github.

I am a founding engineer at Kumo.AI and work on graph machine learning design, its efficiency and scalability, as well as the integration into the Kumo platform as part of our predictive query language. I obtained my PhD at the TU Dortmund University in Machine Learning on graphs. During this time, I created the PyG (PyTorch Geometric) library to easily write and train Graph Neural Networks for a wide range of applications. I am very excited to see PyG being used as a central part of Kumo to drive its machine learning needs, also benefitting PyG and the open-source community in return. I’ve always found the idea of making recent research advancements quickly and widely accessible fascinating. With Kumo, we are finally able to make Graph ML methodologies applicable to non-domain experts as well.

I really enjoy being part of Kumo. I am very grateful for the trust to contribute ideation, prototypes and platform design right from the beginning, and it is amazing to innovate every day with some of the most talented people I’ve ever met. At Kumo, research, collaboration and industrial requirements blend seamlessly.

While away from the computer, I love playing soccer (or as we like to call it: football) and love being outside with our dog. I also play bass in a fun ska band and we tour all over Europe. You can reach me on Twitter and GitHub.