Emma Strubell

Emma Strubell is an NLP researcher whose work links model efficiency, environmental costs, and equity in NLP research. They received an ACL Outstanding Paper Award in 2015, coauthored a fast named-entity recognizer in 2017, and shared EMNLP's 2018 best paper award for LISA. They earned a B.S. in computer science at the University of Maine, completed a Ph.D. at UMass Amherst in 2019, visited Facebook AI Research, and joined CMU's Language Technologies Institute as an assistant professor in 2020. Their 2019 study of NLP training costs won Madrona's inaugural AI Impact Prize in 2021; they led an AI applications theme for Carbon Connect in 2024 and coauthored a 2025 study estimating 493 metric tons of emissions in language-model creation. As of October 2026, they are an associate professor in CMU's LTI and lead SLAB; in June 2026, they coauthored a paper on reducing the inferential load of structured annotation.

Now at: Language Technologies Institute

Timeline

2015

An ACL Outstanding Paper Award

In 2015, they share the ACL Outstanding Paper Award with Luke Vilnis, Kate Silverstein and Andrew McCallum for Learning Dynamic Feature Selection for Fast Sequential Prediction.
2017

Fast entity recognition

In 2017, they coauthor Fast and Accurate Entity Recognition with Iterated Dilated Convolutions, a named-entity recognition system that is more accurate than Bi-LSTM-CRFs while reaching 8x faster test time across whole documents.
2018

LISA and an EMNLP award

In 2018, they coauthor Linguistically-Informed Self-Attention for Semantic Role Labeling. LISA combines multi-head self-attention and multi-task learning across dependency parsing, part-of-speech tagging, predicate detection and semantic role labeling. The paper wins the EMNLP best paper award with four coauthors.
2019–2020

A PhD and move to CMU

They earn a B.S. in computer science from the University of Maine. They complete an M.S. and Ph.D. at UMass Amherst, earning the Ph.D. in 2019 through the Information Extraction and Synthesis Laboratory with Andrew McCallum. They work as a visiting researcher at Facebook AI Research, then join Carnegie Mellon University’s Language Technologies Institute as an assistant professor in 2020.
2019–2021

Energy and policy in NLP

In 2019, they coauthor Energy and Policy Considerations for Deep Learning in NLP, which quantifies the financial and environmental costs of training neural NLP models and recommends ways to reduce costs and improve equity in NLP research and practice. The paper earns Madrona’s inaugural AI Impact Prize in 2021 with Ananya Ganesh and Andrew McCallum.
2024

Carbon Connect’s AI theme

In 2024, they join NSF’s Carbon Connect project as lead of its applications theme on generative AI and virtual spaces. The project plans new standards for carbon accounting in sustainable computing.
2025

Life-cycle costs of language models

In 2025, they coauthor a life-cycle study of language-model creation that accounts for hardware manufacturing, model development and final training runs. It estimates 493 metric tons of carbon emissions, equivalent to the annual power use of about 98 homes in the United States.
As of October 2026

Associate professor and SLAB lead

As of October 2026, they are an associate professor in Carnegie Mellon’s Language Technologies Institute and hold a courtesy faculty appointment in Materials Science and Engineering. They lead SLAB, a research group in natural language processing and machine learning that works on efficient training and inference, data efficiency, expert-domain text, structured prediction, and ethical and policy issues. In June 2026, they coauthor Task Decomposition for Efficient Annotation, which proposes splitting structured annotation into sub-tasks to reduce annotators’ inferential load.