Research Engineer
Google DeepMind
London | |
ramonacomanescu5 at gmail dot com |
I am a Staff Research Engineer at Google DeepMind in London. working on Safety and Fairness of Generative AI models (Gemini).
Previously, I spent 1.5 years as a Machine Learning Engineer working in Community Integrity at Facebook London, building an ML powered identity verification system.
I received my Master of Philosophy (MPhil) in Machine Learning and Machine Intelligence from the University of Cambridge in 2019, being awared a Special Commendation from the Engineering Department for my overall performance. I researched Sum-Product Copula Networks (which combine sum-product networks, a deep probabilistic model with tractable inference, with copulas, a multivariate cumulative distribution function with flexibility for modeling continuous data.) under the supervision of Robert Peharz and Miguel Hernandez Lobato. Previously, I received my Bachelor of Engineering (BEng) in Computer Science from The University of Edinburgh in 2018. During my final year at Edinburgh, I researched gender imbalance in academia using topic modeling and data science methods under the supervision of Adam Lopez.
I am originally from Romania. I moved to the UK in 2014 for university and have since lived in Edinburgh, Cambridge and London. I am a proud dual citizen 🇷🇴🇬🇧.
Ask me about ML, Fairness, good food, good walks and good vibes.
Publications
-
Gaps in the Safety Evaluation of Generative AI
Proceedings of the AAAI/ACM Conference on AI, Ethics, and Society, 2024 -
Holistic Safety and Responsibility Evaluations of Advanced AI Models
Google DeepMind, UK, 2024 -
Gemini 1.5: Unlocking multimodal understanding across millions of tokens of context
Google DeepMind, UK, 2024 -
Gemma: Open Models Based on Gemini Research and Technology
Google DeepMind, UK, 2024 -
Gemini: a family of highly capable multimodal models
Google DeepMind, UK, 2023 -
Doing the right thing for the right reason: Evaluating artificial moral cognition by probing cost insensitivity
Google DeepMind, UK (NeurIPS 2023 MP2 Workshop Poster) -
Improving alignment of dialogue agents via targeted human judgements
Google DeepMind, UK, 2022 -
Melting Pot 2.0
Google DeepMind, UK, 2022 -
Sum-Product Copulas
MPhil Thesis, University of Cambridge, UK, 2019 -
Why is AI "a sea of dudes"? Using data science and NLP methods to understand gender imbalance in a scientific community.
BEng Thesis, The University of Edinburgh, UK, 2018 -
Recognizing emotions in video using multimodal DNN feature fusion
Winner of the Emotion Recognition Track, Proceedings of Grand Challenge and Workshop on Human Multimodal Language (Challenge-HML), 2018 -
DNN multimodal fusion techniques for predicting video sentiment
Proceedings of Grand Challenge and Workshop on Human Multimodal Language (Challenge-HML), 2018
Education
-
MPhil in Machine Learning and Machine Intelligence, 2019
University of Cambridge -
BEng in Computer Science, 2018
The University of Edinburgh