Juan Pino
Research Scientist at Facebook
San Francisco Bay Area United States
Summary
🤩
Rockstar🎓
Top SchoolJuan Pino is a research scientist with 12 years of experience specializing in machine translation, currently driving R&D at Facebook on production neural MT systems that power features like See Translation. He holds a PhD from Cambridge and an MS from Carnegie Mellon, blending deep academic grounding with hands-on engineering to deploy large-scale NMT in production since 2016–2017. An active open-source contributor, Juan has improved PyTorch Translate and fairseq to enable Transformer ONNX export, PyTorch Mobile compatibility, and character encoders—work that broadens deployment to resource-constrained environments. He co-organizes the WeCNLP workshop, reflecting his engagement with both industry and academia, and is particularly focused on improving translation quality and coverage across diverse languages.
12 years of coding experience
1 year of employment as a software developer
Doctor of Philosophy - PhD, Machine Translation, Doctor of Philosophy - PhD, Machine Translation at University of CambridgeÉcole PolytechniqueMaster of Science - MS, Language technologies, Master of Science - MS, Language technologies at Carnegie Mellon University
Stackoverflow
31reputation
7kreached
1answer
0questions
Github Skills (15)
pytorch10
machine-learning10
artificial-intelligence10
nlp10
onnx10
python10
fairseq10
model-optimization10
cprogramming-language9
c-language9
transformers9
transformer9
mapreduce6
hadoop6
java6
Programming languages (9)
Github contributions (5)
pytorch/translate
Mar 2018 - Nov 2020
Translate - a PyTorch Language Library
Role in this project:
ML Engineer
ML EngineerContributions:25 commits, 59 PRs, 4 pushes in 2 years 8 months
Contributions summary:Juan primarily contributed to the PyTorch-based machine translation library. Their commits include modifications to core components like `BatchedBeamSearch`, `NmtDecoder`, and related test files, indicating a focus on improving and optimizing the translation process. The user also addressed issues related to the training pipeline, including fixing update frequencies and ensuring proper testing. Furthermore, the user enhanced the system by adding support for character encoders.
pytorchnlptranslationmachine-learningonnx
facebookresearch/fairseq
Feb 2019 - Oct 2021
Facebook AI Research Sequence-to-Sequence Toolkit written in Python.
Role in this project:ML Engineer
ML EngineerContributions:6 reviews, 78 commits, 25 PRs in 2 years 8 months
Contributions summary:Juan primarily focused on improving the functionality and compatibility of the fairseq library with various deployment environments. Their contributions included adding arguments for ONNX tracing to enable Transformer export, refactoring code for PyTorch Mobile compatibility, and fixing data preparation for speech translation. These changes enabled broader model deployment and optimization for resource-constrained environments. The user also addressed bugs and improved data handling.
pytorchnlpsequencepythontransformer-architecture
Find and Hire Top DevelopersWe’ve analyzed the programming source code of over 60 million software developers on GitHub and scored them by 50,000 skills. Sign-up on Prog,AI to search for software developers.