Abigail See
AI Research Scientist at DeepMind
United Kingdom
Summary
🤩
Rockstar🎓
Top SchoolAbigail See is an AI Research Scientist at DeepMind with a decade of experience focused on natural language processing and sequence-to-sequence models, building on a PhD from Stanford and a mathematics master’s from Cambridge. She has a strong track record in production-ready research: contributions to influential open-source projects include robustness fixes and beam search improvements for the widely cited pointer-generator summarization code and work on Facebook’s ParlAI dialogue framework. Her internships at Google Brain and Facebook AI produced practical advances in long-document summarization, interpretability, and controllable text generation, reflecting a blend of theoretical rigor and engineering pragmatism. Based in the UK, she combines deep academic pedigree with hands-on ML engineering to improve model stability and real-world dialogue behavior.
10 years of coding experience
Master’s Degree, Mathematics, Distinction, Master’s Degree, Mathematics, Distinction at University of CambridgeDoctor of Philosophy - PhD, Computer Science, Doctor of Philosophy - PhD, Computer Science at Stanford University
English, French
Stackoverflow
1reputation
0reached
0answers
0questions
Github Skills (17)
beam-search10
pytorch10
python10
text-summarization10
machine-learning10
trainings10
tensorflow10
natural-language-processing10
sequence-to-sequence10
nlp10
modeling10
data-processing9
ai9
model-optimization8
git7
Programming languages (6)
Github contributions (5)
abisee/pointer-generator
Apr 2017 - Apr 2019
Code for the ACL 2017 paper "Get To The Point: Summarization with Pointer-Generator Networks"
Role in this project:
ML Engineer
ML EngineerContributions:29 commits, 2 PRs, 26 pushes in 2 years
Contributions summary:Abigail primarily focused on improving the codebase related to the pointer-generator network for text summarization. Their contributions included fixing a bug in beam search, addressing issues with empty article texts, and implementing various changes to handle potential NaN (Not a Number) values in the model training process. These changes involved modifying the loss calculation, incorporating padding masks, and adding features for debugging and model restoration, indicating a focus on model robustness and stability.
nlppointer-generatorpointpointersummarization
facebookresearch/ParlAI
Nov 2018 - May 2019
A framework for training and evaluating AI models on a variety of openly available dialogue datasets.
Role in this project:ML Engineer
ML EngineerContributions:7 commits, 7 PRs, 7 pushes in 6 months
Contributions summary:Abigail contributed to the `parlai` repository, which focuses on dialogue AI models. Their commits included fixes to the beam search algorithm, a core component for generating responses in sequence-to-sequence models. Furthermore, the user added functionality to the `TorchAgent` class with flags for historical context and token manipulation, likely improving model behavior. They also made changes related to loss functions in the `Seq2seqAgent` and addressed a backward compatibility issue with PyTorch, demonstrating a focus on model efficiency and maintenance.
ai
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.