Yih-dar Shieh
ML Engineer at Hugging Face
Paris, Ile-de-France
Summary
🤩
Rockstar🎓
Top SchoolYih-dar Shieh is an ML Engineer based in Paris with 10 years of experience bridging deep mathematical research and production-grade machine learning. A PhD in mathematics and an engineering degree in computer science, he has shipped large ML and NLP improvements in industry—lifting model performance from 0.78 to 0.95 and making rule-based systems >100× faster with ~10× less RAM. At Hugging Face he contributes to core open-source libraries (Transformers and Diffusers), improving TF model serving and key diffusion components used broadly across the community. He is an active Kaggle competitor and TPU Star notebook author, demonstrating both practical engineering at scale (models for 1M+ students) and reproducible research. Comfortable across TensorFlow, PyTorch, TFLite and cloud platforms, he blends rigorous academic problem-solving with pragmatic optimization for production. An underappreciated strength: his background in p‑adic algorithms and algebraic curves gives him uncommon expertise in formal algorithmic thinking that he applies to model and system-level optimizations.
10 years of coding experience
3 years of employment as a software developer
Diplôme d'ingénieur, Informatique, Diplôme d'ingénieur, Informatique at Polytech'MarseilleDoctorat, Mathématiques, Mention très honorable, Doctorat, Mathématiques, Mention très honorable at Aix-Marseille Université
Chinese, English, French
Github Skills (22)
pytorch10
integrate10
python10
diffusion-models10
machine-learning10
integrations10
diffusers10
diffusion-probabilistic-models10
diffusion-probabilistic10
deep-learning10
image-generation10
tensorflow10
resnet10
data-integration10
stable-diffusion9
Programming languages (8)
Github contributions (5)
huggingface/transformers
Mar 2021 - Jan 2023
🤗 Transformers: State-of-the-art Machine Learning for Pytorch, TensorFlow, and JAX.
Role in this project:
ML Engineer
ML EngineerContributions:3433 reviews, 626 commits, 2232 PRs in 1 year 10 months
Contributions summary:Yih-dar's contributions primarily focused on updating and modifying TensorFlow models within the Hugging Face Transformers library, specifically for serving output in several TF-based models. The contributions include changes to the code for TF models like TFLEDModel, TFT5Model, and TFWav2Vec2Model, enhancing their compatibility and functionality. The user also applied the same changes to the TFVisionEncoderDecoderModel and TF T5ForConditionalGeneration model to enhance their serving capabilities.
pythonbertspeech-recognitionstate-of-the-artflax
huggingface/diffusers
Aug 2022 - Sep 2022
🤗 Diffusers: State-of-the-art diffusion models for image, video, and audio generation in PyTorch and FLAX.
Role in this project:ML Engineer
ML EngineerContributions:28 reviews, 23 commits, 12 PRs in 1 month
Contributions summary:Yih-dar primarily focused on improving and maintaining the code related to the diffusion models, a key element of the Hugging Face diffusers repository. Their contributions involved fixing and refining components, specifically related to upsampling, downsampling, and cross-attention mechanisms within the model's architecture. They corrected logical errors and ensured the correct functionality of critical model components. Their work demonstrates a good understanding of the underlying diffusion model implementation.
pytorchartdeep-learningimage2imagestate-of-the-art
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.