Han Xiao
Vice President Of AI at Elastic
Mountain View, California, United States
Summary
🤩
Rockstar🎓
Top SchoolHan Xiao is a Vice President of AI at Elastic and a founder-turned-exit leader who built Jina AI and led it through acquisition, bringing 11 years of experience in productionizing search and multimodal AI. With a PhD from TUM and prior roles leading search at Tencent and research at Zalando, he blends deep academic rigor with product-driven engineering. He is the creator of the widely cited Fashion-MNIST dataset and has hands-on open-source contributions across Jina, DocArray, and several multimodal model pipelines, focusing on backend, gRPC, and model integration. Known for shipping robust inference and embedding infrastructure, he repeatedly bridges research prototypes to cloud-native deployments. Based in Mountain View, he pairs founder-level strategic vision with detailed engineering work such as type-safe document models and compression-aware client refactors.
11 years of coding experience
12 years of employment as a software developer
Resident Researcher Computer Science, Resident Researcher Computer Science at National Taiwan UniversityDoctor of Philosophy (Dr.rer.nat.) Computer Science, Doctor of Philosophy (Dr.rer.nat.) Computer Science at Technical University of Munich
Stackoverflow
71reputation
1kreached
2answers
1question
Github Skills (70)
python10
image-processing10
azure-functions10
firebase-cloud-functions10
semantic-search10
prompt-engineering10
deep-learning10
image-generation10
trainings10
pdf-document10
api10
generative-art10
typescripts10
grpc10
puppeteer10
Programming languages (9)
Github contributions (5)
jina-ai/discoart
Jun 2022 - Aug 2022
🪩 Create Disco Diffusion artworks in one line
Role in this project:
Full-stack Developer
Full-stack DeveloperContributions:3 reviews, 336 commits, 181 PRs in 1 month
Contributions summary:Han's initial commit focuses on setting up the foundational aspects of the "discoart" project, including essential helper functions, a runner script, and the initialization of a create function. Further commits reveal a focus on model loading, specifically for open-clip models, and implementation of features like text prompts and initial images. The user also appears to have been involved in debugging and fixing display-related issues within the project.
disco-diffusioncreative-aidiffusioncreative-artlatent-diffusion
jina-ai/dalle-flow
Apr 2022 - Jul 2022
🌊 A Human-in-the-Loop workflow for creating HD images from text
Role in this project:ML Engineer
ML EngineerContributions:4 reviews, 230 commits, 22 PRs in 2 months
Contributions summary:Han contributed to the development of image generation capabilities within the dalle-flow repository. Their work included modifying the dalle-mini model implementation, integrating GLID3-XL for diffusion, and implementing upscaling functionalities. They also added timing metrics for performance analysis. The commits suggest a focus on refining the image generation pipeline for the project.
pythonstablediffusiontext-to-imagedalle-megaglid3
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.