Santiago Castro
Research Scientist at Netflix
San Francisco, California, United States
Summary
🤩
Rockstar🎓
Top SchoolSantiago Castro is a Research Scientist at Netflix with 12 years of experience bridging academic rigor and industry impact in vision-and-language and video representation learning. He holds a PhD in Computer Science from the University of Michigan and a track record of multiple Netflix, Google, and Adobe research internships focused on compositional generalization and language-guided video representations. Santiago combines deep research skills with pragmatic engineering—contributing bug fixes, refactors, and performance-focused changes to major open-source projects like Hugging Face datasets, Transformers, and Scenic. Earlier in his career he founded and led ML efforts at Xmartlabs, creating Bender, an open-source framework for real-time neural networks on iOS that presaged later platform offerings. Based in San Francisco, he brings both production-grade code hygiene and a history of improving documentation and reliability across NLP and CV toolchains. He’s equally comfortable publishing novel research and rolling up his sleeves to maintain large community libraries.
12 years of coding experience
11 years of employment as a software developer
Doctor of Philosophy - PhD Computer Science, Doctor of Philosophy - PhD Computer Science at University of MichiganHigh School Diploma Engineering, High School Diploma Engineering at Escuela y Liceo Elbio FernándezBachelor of Engineering - BE Computer Science, Bachelor of Engineering - BE Computer Science at Universidad de la República
Spanish, English
Stackoverflow
5,940reputation
423kreached
27answers
7questions
android
top-5%
kotlin
top-5%
okhttp
top-5%
cuda
top-1%
gcc
top-5%
Github Skills (70)
transformers10
python10
flax10
datasets10
metal10
deep-learning10
apple210
ml10
computer-engineering10
neural-network10
computer-vision10
transformer10
swift10
csv10
nlp10
Programming languages (34)
Github contributions (5)
huggingface/transformers
Sep 2019 - Sep 2022
🤗 Transformers: the model-definition framework for state-of-the-art machine learning models in text, vision, audio, and multimodal models, for both inference and training.
Role in this project:
ML Engineer
ML EngineerContributions:4 reviews, 19 commits, 22 PRs in 3 years
Contributions summary:Santiago made several contributions focused on improving the Hugging Face Transformers library, particularly in the area of machine learning model implementation and optimization. Their work includes removing unnecessary code, such as redundant use of `FusedLayerNorm` in various models like BERT and XLNet, and correcting minor documentation errors. They also added support for the null answer in the `QuestionAnsweringPipeline` and fixed typos to improve the library's usability and code quality. Furthermore, the user addressed various doc errors and typos, enhancing the overall clarity of the codebase.
audioinferencemachine-learning-modelsmultimodaltransformers
xmartlabs/MetalPerformanceShadersProxy
Jul 2017 - Aug 2017
A proxy for MetalPerformanceShaders which takes to a stub on a simulator and to the real implementation on iOS devices.
Contributions:4 releases, 37 commits, 4 PRs in 20 days
iosmetal-performance-shadersproxymetalmetalkit
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.