Andreas Grivas
Postdoctoral Machine Learning Researcher at The University of Edinburgh
City of Edinburgh, Scotland, United Kingdom
Summary
🤩
Rockstar🎓
Top SchoolAndreas Grivas is a postdoctoral machine learning researcher with 11 years of experience applying geometric insights to make text representations more expressive, reliable, and efficient. Based at the University of Edinburgh, he combines deep mathematical understanding of LLMs with hands-on engineering—leading teams, coding novel loss functions in PyTorch, and running large-scale training on Slurm and Kubernetes. His work has accelerated LLM generation by over 5× and translated into practical tools that help journalists, clinicians, and portfolio managers extract actionable insights from text. An active open-source contributor, he has improved core NLP infrastructure through contributions to spaCy, enhancing interactive representations and robust unicode handling. He enjoys communicating model structure through visualisations and has a track record of finding and fixing real-world data and annotation issues using FAISS and principled analysis.
11 years of coding experience
8 years of employment as a software developer
Doctor of Philosophy - PhD Artificial Intelligence, Doctor of Philosophy - PhD Artificial Intelligence at The University of EdinburghBachelor's degree Informatics and Telematics, Bachelor's degree Informatics and Telematics at Harokopio University of Athens
English, Greek
Stackoverflow
367reputation
92kreached
7answers
0questions
Github Skills (17)
python10
natural-language-processing10
cython10
nlp10
tokenize9
tokenizer9
data-structures8
data-structure8
testing7
bibtex6
boost-python6
boost6
tmux6
pyqt6
msword6
Programming languages (8)
Github contributions (5)
explosion/spaCy
Oct 2015 - Nov 2015
💫 Industrial-strength Natural Language Processing (NLP) in Python
Role in this project:
Back-end Developer
Back-end DeveloperContributions:5 commits, 8 PRs, 14 comments in 15 days
Contributions summary:Andreas primarily contributed to the core functionality of the spaCy library by adding and modifying methods like `__repr__`, `__str__`, and `__bytes__` for the `Doc`, `Token`, and `Span` objects to improve their display in interactive environments. They fixed issues related to unicode printing across the library and implemented changes in the test suite for print and representation functionality. The user's work also involved modifying span merging and handling for correct indices, ensuring accurate representation of merged spans within the document.
natural-language-processingpythondata-sciencemachine-learningcython
pan-webis-de/pangram
Aug 2015 - Nov 2015
Contributions:27 commits, 14 pushes, 3 branches in 2 months
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.