Sign In

Guillaume Becquin

Senior Research Engineer

Guillaume Becquin is a Senior Research Engineer at Bloomberg LP, specializing in Natural Language Processing (NLP) within the company's Artificial Intelligence (AI) Engineering department. He is involved in developing advanced AI solutions that enhance search, discovery, and workflow for Bloomberg's financial data products. His work includes designing and evaluating NLP models, contributing to research, and collaborating on production systems to improve the quality of data and analytics derived from unstructured information.125

Becquin holds a degree from the École nationale supérieure de l'Aéronautique et de l'Espace, indicating a strong technical background. He has been recognized for his contributions to AI research, including co-authoring papers presented at significant conferences such as EMNLP 2023, focusing on innovative approaches to paraphrase generation.345 He is based in London and has built a professional network of over 500 connections on LinkedIn, reflecting his active engagement in the field.67

Highlights

Nov 13 · linkedin

Natural Language Processing would not be where it is today without the power of Open Source Software. The collective knowledge and input of thousands of contributors enabled the development of libraries that are powering many research and industry NLP projects. I am thankful to the contributors of libraries such as NLTK, SpaCy, AllenNLP or Transformers.

A few months ago, I started a personal project in an attempt to give something back to the community: a Rust port of Hugging Face's Transformers library (rust-bert). Today I am humbled and excited to share that an article about this project has been accepted for publication at the NLP-OSS workshop hosted at #emnlp2020.

The article is already available at https://lnkd.in/dCSWFhi and will be presented during the NLP-OSS workshop next week. Looking forward to engage the discussion there, and learn more about what's next in #nlposs

#nlp #opensource #opensourcesoftware #rustlang #deeplearning

Jun 1 · linkedin

Tokenizers are often taken for granted by NLP practioners but have been fundamental in driving the field's progress over the past few years. The SentencePiece Unigram model is a modern example relying on a statistical approach to text segmentation. However, I felt there were limited code examples beyond the reference C++ library that were easily readable and usable in real life settings.

Attempting to fill this gap, I wrote a short article walking through the Unigram encoding algorithm illustrated with two Rust implementations. These fit in less than 150 lines of code, and the fastest offers a performance similar to the reference C++ library.

The article is available at https://lnkd.in/gyfg49P . I hope you find it useful! #rust #nlp #sentencepiece #tokenizer #datascience

Related Questions

What specific projects has Guillaume Becquin worked on at Bloomberg LP?
What are Guillaume Becquin's main research interests in NLP?
How long has Guillaume Becquin been with Bloomberg LP?
What are some notable achievements of Guillaume Becquin in the field of AI?
Does Guillaume Becquin publish research papers or articles?
Guillaume Becquin
Guillaume Becquin, photo 1
Guillaume Becquin, photo 2
Add to my network

Location

London, England, United Kingdom