Brian Thompson

<firstname>@alumni.caltech.edu

I am currently taking a sabbatical to pursue some travel and personal projects. If you have consulting work or a particularly compelling full-time opportunity that you think I would be well suited for, please reach out using the email above.

About Me

I was previously a Senior Applied Scientist in Amazon's Artificial General Intelligence (AGI) org, where I helped launch the Nova AI models. Before that, I worked at Apple, Johns Hopkins University (where I also completed my PhD), MIT Lincoln Laboratory, and Rincon Research Corporation, on topics including large language model (LLM) training, machine translation (MT), automatic dubbing, text-to-speech (TTS), data curation and filtering, automatic and human evaluation, multilingual modeling, paraphrasing, cross-language information retrieval, domain adaptation, and digital signal processing.

My work exploring the impact of machine translation on the web has been covered by Politico, The Atlantic, Slator, Vice, TechInsider, Futurism, and Forbes.

Open Source Projects

I developed the Vecalign sentence alignment toolkit for the ParaCrawl parallel data acquisition project. Vecalign is an accurate sentence alignment algorithm based on multilingual sentence embeddings which is linear in complexity with respect to the number of sentences being aligned. In conjunction with a multilingual sentence embedding model like LASER or LaBSE, Vecalign makes it easy to perform sentence alignment in about 100 languages (i.e. 100^2 language pairs), without the need for a machine translation system or lexicon. At the time of writing, Vecalign has the best reported performance on the test set released with Bleualign.

I also developed Prism, an automatic MT metric which uses a sequence-to-sequence paraphraser to score MT system outputs conditioned on their respective human references. Prism uses a multilingual neural MT model as a zero-shot paraphraser, which eliminates the need for synthetic paraphrase data and results in a single model which works in many languages. To the best of my knowledge, the Prism model was the first large open-source multilingual translation model. The filtering code to preprocess the data used to train Prism is general enough to use for any MT training and is released here.

Academic Service

I co-organized the IWSLT (2025, 2024, 2023), WMT metrics (2025, 2024, 2023), and WMT data curation (2023) shared tasks.

I have also served as an area chair and/or reviewer for numerous organizations including the Association for Computational Linguistics (ACL); Association for the Advancement of Artificial Intelligence (AAAI); Conference on Machine Translation (WMT); IEEE Transactions on Audio, Speech, and Language processing; International Conference on Spoken Language Translation (IWSLT); and INTERSPEECH.

Education

The Johns Hopkins University

California Institute of Technology

Rose-Hulman Institute of Technology

Publications

Note: Google Scholar may be more up-to-date.