Leading the development of Large Language Models for Enriching Amazon's Global Product Catalog.- Training Large Language Models for Amazon Worldwide Catalog.

Leading the development of Large Language Models for Amazon's Global Product Catalog.- I have been working at the forefront of integrating Generative AI into Amazon Catalog Enrichment processes. My pioneering work includes the creation of the first Generative-LLM, called SAGE, designed for multi-lingual attribute value predictions from unstructured text and product images. The first version of the SAGE model went live in April 2022. - I invented a novel Diversified Sampling Algorithm, which has dramatically reduced the need for human labeling in training Generative AI models. - I introduced the Zero-Shot Expansion strategy for SAGE, utilizing a novel graph-based method that has led to a 9-fold decrease in human labeling costs. I have also developed SAGE-L, a variant of SAGE that transcends specific marketplaces, thereby significantly speeding up ML coverage of the Catalog without incurring additional human-labeling costs. - I invented a custom Adaptive Batching Algorithm has resulted in a 3x decrease in inference costs for operating SAGE models in a production environment.The impact of SAGE models in Amazon has been profound. 3 months after first launch in production, my models became the primary source of attribute values in Amazon's Worldwide Catalog, aside from sellers. They have been causally linked to very significant improvement in customer experience, as well as, very significant revenue increase for the catalog.

Research in Computational Genomics / Cancer Informatics: -- Computational and Mathematical Models of Cancer Evolution.

Research in Machine Learning / Artificial Intelligence: -- Research Output Highlights : Top-N Recommendation Algorithms: RecWalk, PerDIF. (Received 1 Best Paper Award and 1 Nomination for Best Paper Award) -- Mining and Learning over Graphs: Adaptive Diffusions (Received 1 Best Paper Award). -- Development of Seminar Series on Emerging Data Science Methodology -- Participation in the Funding Efforts of the Digital Technology Center. Funding Highlights: Co-Principal Investigator in an awarded NSF Grant: NSF award number 1901134.

ML Research with an emphasis on Graph Learning and Recommender Systems.

Junior Research Associate -- Research Output Highlights: Link Analysis Algorithms: Notably NCDawareRank (NCDawareRank has since been adopted by Start-Ups in the Blockchain community as their core network centrality measure). Design of Novel Recommendation Algorithms: Notably NCDREC, Hierarchical Itemspace Rank (the central focus of which was to alleviate sparsity-related degradation of collaborative filtering performance (e.g., Cold-Start and Long-Tail problems). -- Software Engineer in several EU- and Nationally-funded research and development projects including "Digital Trust", "E_STEP", and "Digital School".