Michael Günther
Principal AI Research Engineer I at Elastic
Berlin, Germany
Summary
🤩
Rockstar🎓
Top SchoolMichael Günther is a Principal AI Research Engineer with 12 years of experience building multimodal AI systems and integrating deep learning into production-grade software. With a Ph.D. from TU Dresden on embedding NLP into relational databases, he blends rigorous research with practical engineering at companies like Jina AI and Elastic. He has driven backend and ML work on notable open-source projects such as DocArray and Finetuner—extending device support across frameworks and improving multimodal embedding evaluation. Based in Berlin, he excels at bridging model research and scalable infrastructure, often tackling device portability and evaluation plumbing that most researchers leave to engineers.
12 years of coding experience
11 years of employment as a software developer
Ph.D. (Dr.-Ing.), Computer Science, Ph.D. (Dr.-Ing.), Computer Science at Technische Universität Dresden
Stackoverflow
1reputation
0reached
0answers
0questions
Github Skills (20)
pytorch10
python10
evaluation10
machine-learning10
keras10
fine-tuning10
paddlepaddle10
transfer-learning10
metric10
nearest-neighbors9
testing9
data-structure9
pytest9
bert9
nearest-neighbor-search9
Programming languages (7)
Github contributions (5)
jina-ai/finetuner
Feb 2022 - Jan 2023
:dart: Task-oriented embedding tuning for BERT, CLIP, etc.
Role in this project:
Full-stack Developer
Full-stack DeveloperContributions:5 releases, 161 reviews, 46 commits in 11 months
Contributions summary:Michael made significant contributions related to device management and framework integration within the finetuner library. They added device options for PaddlePaddle, PyTorch, and Keras, which involved implementing framework-specific device retrieval and mapping functions. This work extended the existing functionality to support different computing environments, specifically focusing on enabling the model to run on CUDA and CPU devices. Furthermore, the user worked on the integration of the tailor module to the project, which likely required adaptation and modification to different models within the project's infrastructure.
nlptriplet-lossfinetuningsiamese-networkjina
docarray/docarray
Aug 2022 - Dec 2022
Represent, send, store and search multimodal data
Role in this project:Back-end Developer & ML Engineer
Back-end Developer & ML EngineerContributions:23 reviews, 10 commits, 14 PRs in 4 months
Contributions summary:Michael primarily contributed to the back-end of the DocArray project, focusing on refactoring and implementing new features related to evaluation metrics and the `embed_and_evaluate` function. Their work involved modifying existing code, introducing new functionalities, and adding tests to ensure the quality and reliability of the project. The user demonstrated expertise in integrating with the project's existing architecture, which revolves around represention, storage, and searching multimodal data.
protobuffeature-storevectorunstructured-dataqdrant
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.