Scholar
Valentina Pyatkin
Google Scholar ID: E9EgKkMAAAAJ
Allen Institute for AI & University of Washington
Citations & Impact
Citations
3,026
H-index
19
i10-index
29
Publications
20
Co-authors
129
list available
Publications
Cited
0
Resume (English only)
Academic Achievements
- ACL Outstanding Paper Award.
- ACL Best Theme Paper Award.
- AI2 Outstanding Intern of the Year Award.
- Two paper awards at ACL 2024.
- Two papers accepted to ICML 2025.
- Tutorial Chair for EMNLP 2025.
- Internal Communication Chair for ACL 2024.
- Invited talks at top venues and institutions including ICML, NeurIPS, ACL, Stanford, Harvard, Oxford, NVIDIA, and EPFL.
Research Experience
- Postdoctoral researcher at the Allen Institute for AI and the University of Washington, advised by Prof. Hanna Hajishirzi.
- Research intern at Google.
- Core contributor to the Tulu and Open-Instruct projects, developing post-training pipelines involving supervised fine-tuning, direct preference optimization, and reinforcement learning with verifiable rewards.
- Contributed to open-source LLM projects OLMo and OLMo2.
- Co-organized workshops including SoLaR (Socially Responsible Language Modelling Research) and UnImplicit.
Education
- PhD in Computer Science from Bar Ilan University, NLP Lab, supervised by Prof. Ido Dagan and Prof. Reut Tsarfaty.
- MSc from the University of Edinburgh.
- BA from the University of Zurich.
- Visited UW NLP as a PhD student under the supervision of Prof. Yejin Choi.
- Completed two research internships at the Allen Institute for AI.
Background
- Currently a Postdoctoral Researcher (and Young Investigator) at the Allen Institute for AI and the University of Washington.
- Research focuses on developing generative AI that is contextually robust, responsible, and open.
- Particularly interested in extending language model capabilities through post-training and adaptation.
- Contributed to widely-used benchmarks such as RewardBench.
- Research areas include: open science of LLMs and post-training, steerability, underspecification, and precise contextual response.
