Leonardo Ribeiro

About

Hello!

I am a machine learning researcher working on large language models, focusing on how to make them more reliable, controllable, and aligned with human goals.

I’m currently a Principal Applied Scientist at Microsoft, where I work on Responsible AI and alignment, developing techniques to improve the behavior, reasoning, and safety of large-scale models.

From 2022 to early 2026, I was a Senior Applied Scientist at Amazon, most recently in Amazon AGI working on large-scale post-training for foundation models — improving reasoning, controllability, and product quality across real-world applications. My work there spanned deep research / agentic search systems, RAG, controllable generation, and faithful summarization, including projects like GaRAGe (a benchmark with grounding annotations for RAG evaluation), FactGraph (semantic-graph-based factuality evaluation for summarization), and controllable-readability summarization.

I completed my Ph.D. in computer science at the Technical University of Darmstadt in the UKP Lab, advised by Iryna Gurevych. My thesis focused on structured representations and graph-based methods for natural language generation, including AMR-to-text and knowledge-graph-to-text generation. Earlier, I worked on graph representation learning with Daniel R. Figueiredo at UFRJ, including struc2vec (KDD 2017).

My work has been published at ACL, EMNLP, NAACL, KDD, and TACL. I support the research community as Senior Area Chair (NAACL 2025), Tutorial Chair (ACL 2024), and Area Chair across ACL, EACL, EMNLP, and COLING.