I’m an Applied Scientist at the Generative AI Innovation Center at Amazon Web Services. My work sits at the intersection of AI research and real-world deployment: I develop evaluation methodologies and multi-agent architectures that enable enterprises across critical industries to reliably deploy large language models for mission-critical processes.

Research Interests

• AI Evaluation: Designing rigorous methods for assessing generative AI systems, from structured output evaluation for information extraction to confidence calibration in agentic environments
• Multi-Agent Systems: Designing and deploying systems that coordinate multiple specialized AI agents to solve complex enterprise problems across industries
• Synthetic Data for Evaluation: Establishing validity frameworks for synthetic evaluation data, addressing the fundamental bottleneck of data access in regulated industries
• LLM Reliability & Calibration: Two active research threads: (1) quantifying structural consistency in LLM logical reasoning as a measure of uncertainty (ICLR 2026 Best Paper), and (2) pass@k calibration, studying whether LLMs can predict their own marginal gains from additional sampling to enable efficient inference-time compute allocation

Recent Highlights

• Best Paper Award at ICLR 2026 Workshop on Logical Reasoning of Large Language Models for work on quantifying consistency in LLM logical reasoning
• Co-developed stickler, an open-source evaluation framework for structured LLM outputs (7,000+ monthly PyPI downloads). I’m one of the main contributors to the hierarchical aggregate metrics system for evaluating complex nested structures at arbitrary depth.
• Main author of Evaluating Intelligent Document Processing Solutions in the GenAI Atlas
• Organized the AMLC 2025 Workshop on Evaluation Paradigms for GenAI (83 paper submissions, 100+ attendees) and organizing the AMLC 2026 Workshop on Test-Time Scaling: Algorithms, Systems, and Evaluation for Inference-Time Compute in the Era of Reasoning Models
• AI solutions deployed across 9+ industry sectors: energy, pharmaceutical manufacturing, financial services, media, legal, HR, and consumer retail

Background

I completed my PhD in Computational Linguistics at the University of Illinois at Urbana-Champaign (alongside an MS in Statistics), where I worked with Prof. Lane Schwartz on computational methods for low-resource languages. My research spanned multilingual language modeling (TACL 2021), neural approaches to polysynthetic language modeling at JSALT 2019, and work supported by NSF. My dissertation brought much of this work together.

I’ve been at Amazon since 2022, previously working on LLM-based content generation for Alexa Kids and NLU for Fire TV before joining the GenAI Innovation Center.

Feel free to reach out at hayley.uiuc@gmail.com

---

Selected Publications

• [Best Paper Award] Quantifying Consistency in LLM Logical Reasoning via Structural Uncertainty
  Baishali Chaudhury, Mengdie Flora Wang, Hyunji Hayley Park, Rahul Ghosh, Sungmin Hong, Jae Oh Woo
  ICLR 2026 Workshop on Logical Reasoning of Large Language Models, 2026

• Halliburton Enhances Seismic Workflow Creation with Amazon Bedrock and Generative AI
  Yuan Tian, Arun Ramanathan, Baishali Chaudhury, Gan Luan, Haochen Xie, Jared Kramer, Hayley Park, Di Wu
  AWS Machine Learning Blog, 2026. A version of this work was also co-published with Halliburton in the EAGE 2026 Proceedings.

• Beyond Binary Metrics: A Structured Evaluation Framework for Information Extraction
  Spencer Romo, Ayushi Haria, Hyunji Hayley Park, Sujitha Martin, Piyush Vyas, Jared Kramer
  AMLC 2025 Workshop on Evaluation Paradigms for GenAI, 2025

• Efficient Classification of Long Documents Using Transformers
  Hyunji Hayley Park, Yogarshi Vyas, Kashif Shah
  Association for Computational Linguistics (ACL), 2022

• Expanding Universal Dependencies for Polysynthetic Languages: A Case of St. Lawrence Island Yupik
  Hyunji Hayley Park, Lane Schwartz, Francis M. Tyers
  Workshop on NLP for Indigenous Languages of the Americas (AmericasNLP), NAACL, 2021

• Morphology Matters: A Multilingual Language Modeling Analysis
  Hyunji Hayley Park, Katherine J. Zhang, Coleman Haley, Kenneth Steimel, Han Liu, Lane Schwartz
  Transactions of the Association for Computational Linguistics (TACL), 2021

• Neural Polysynthetic Language Modelling
  Lane Schwartz, Francis Tyers, …, Hyunji Hayley Park, et al.
  Final Report of the Neural Polysynthetic Language Modelling Team at the 2019 Frederick Jelinek Memorial Summer Workshop, 2020

For a full list, see my Google Scholar.