Home
Short bio
Vassilis N. Ioannidis is a Senior Applied Scientist at Amazon AI. He has a Ph.D. degree in “Robust Deep Learning on Graphs” from the University of Minnesota and was awarded the Doctoral Dissertation Fellowship. Vassilis has published 50+ conference and journal papers, mentored 25+ scientists and delivered 18 tutorials and workshops at conferences and universities. He worked from June to Dec. 2019 at MERL as an ML scientist in graph representation learning. Since Feb. 2020, he has been working at Amazon, where he develops GNN and NLP solutions. He leads the research and development in the intersection of GraphML and NLP with applications in AWS and Amazon. He developed Neptune ML that is a machine learning service over graph databases deployed in AWS using GNNs. He led large scale training of foundational GNN and NLP models for Amazon Search AI projects in information retrieval, generative AI, recommendation and abuse detection resulting to over 1B dollars revenue. The work was open-sourced in GraphStorm, an open source framework combining GNNs and LLMs. He has led teams of scientists to launch new Amazon Search features based on graph and language technologies to solve the upper funnel shopping stage. Recently, he leads graph-RAG a cross-org initiative on enhancing RAG systems for structured data.
Bio sketch
Vassilis N. Ioannidis is a Senior Applied Scientist at Amazon Search AI with a Ph.D. in "Robust Deep Learning on Graphs" from the University of Minnesota. He's published 50+ papers, mentored 25+ scientists, and led projects combining GraphML and NLP for AWS and Amazon. He's a key contributor to Neptune ML and GraphStorm and led teams in developing new Amazon Search features based on graph and language technologies. Recently, he leads graph-RAG a cross-org initiative on enhancing RAG systems for structured data.
Email: vassilisnioannidis@gmail.com
Contact me for my latest CV
Welcome!
Since Feb. 2020 I work at the AWS AI Research and Education as a Senior Applied Scientist in the AWS Graph ML/DGL team. We design, implement and apply graph neural networks and graph machine learning to solve customers problems in AWS and Amazon. I have worked in the Neptune ML services that employs GNNs a variety of problems for customers of the Amazon Neptune database. I also work in retrieval, fraud detection and recommendation problems at Amazon by applying distributed GNNs and distributed language models at trillion edge graphs. Besides these project we also perform high impact research, and publish at top-tier conferences. I have mentored more than 30 scientists performing fundamental core modeling, system and application-focused research in GNNs. Please reach out if you are looking an internship in the area of GNNs and/or language models.
I obtained my Doctorate degree in the Department of Electrical and Computer Engineering at the University of Minnesota (UMN) under the supervision of Prof. Georgios B. Giannakis on Aug. 2020. My research leverages advances in deep learning on graphs, optimization, data science, tensor decomposition, and machine learning to address robust learning tasks over large-scale dynamic networks. Research outcomes include scalable and online algorithms for predicting time-series on graphs, coupled tensor and graph factorization techniques for recommender systems and community detection, and graph neural network architectures. I obtained my (5-year) Diploma from the School of Electrical and Computer Engineering of the National Technical University of Athens in 2015 with Major in Computer Science. I also obtained my M.Sc. degree from UMN in 2017. From Sep. 2014 to Aug. 2015 I worked at Oracle as a Software Engineer, in the design and development of a Machine to Machine platform for Vodafone Group. From June 2019 to Dec. 2019 I worked at Mitshubishi Electric Research Labs as a Research Scientist in the design of novel Graph Convolutional Networks with application to 3D point cloud processing. We also analyzed key stability properties of methods for deep learning on graphs. Find out more in our ICLR publication.