Isabelle Ng

Computer Science • Data Science • Music — UC Berkeley

I am a final year undergraduate student at UC Berkeley interested in natural language processing. My most recent work involves designing and evaluating agentic LLM systems that combine evolutionary search with inference-time refinement, with applications to text-to-SQL. I am currently seeking PhD advisors and applying to PhD programs for the next cycle. I’ll be at NeurIPS 2025 - happy to chat about research and potential advising fits.

NLP Agentic LLMs AI Engineering Machine Learning AI Safety
GitHub LinkedIn Email OpenReview

Publications & Preprints

2025

MultiGA: Leveraging Multi-Source Seeding in Genetic Algorithms

Isabelle Diana May-Xin Ng, Tharindu Cyril Weerasooriya, Haitao Zhu, Wei Wei • NeurIPS 2025 Efficient Reasoning Workshop (to appear)

PDF (soon) Poster (soon)
2025

ProRefine: Inference-Time Prompt Refinement with Textual Feedback (preprint)

Deepak Pandita, Tharindu Cyril Weerasooriya, Ankit Parag Shah, Isabelle Diana May-Xin Ng, Christopher M. Homan, Wei Wei • arXiv: 2506.05305

PDF

Research & Industry Experience

GenAI Research Engineering Intern

Accenture, Center for Advanced AI · Mountain View, CA · Jun 2025 – Aug 2025

  • Led research and development of MultiGA, a novel evolutionary prompting framework for multi-LLM collaboration on reasoning tasks such as text-to-SQL and trip planning.
  • Developed an experimental pipeline enabling automated agentic experiments with remote metric collection and evaluation.
  • First author of MultiGA, to appear at NeurIPS 2025 Efficient Reasoning Workshop and in progress for ACL ARR 2026 submission.

NLP Student Researcher

UC Berkeley School of Information · Berkeley, CA · Jan 2025 – May 2025

  • Analyzed 1,500 legal cases using LLMs to study judicial reasoning around Section 230 under UCB/Stanford JD-PhD mentorship.
  • Identified interpretive patterns with implications for legislative reform; project recognized with Best Poster at UC Berkeley’s Research Symposium.
  • See Berkeley I School news coverage here.

Data Engineering Consultant

Mattel, Inc. · Remote · Sep 2024 – Jan 2025

  • Assessed Mattel’s enterprise data ecosystem and designed a scalable ESG data pipeline supporting sustainability initiatives.
  • Conducted stakeholder interviews and identified infrastructure discrepancies across divisions.
  • Developed recommendations to centralize analytics workflows in Google BigQuery and improve governance and access protocols.

Machine Learning Engineering Intern

H2O.ai · Mountain View, CA · Jun 2024 – Aug 2024

  • Built a Retrieval-Augmented Generation (RAG) chatbot to assist employees with technical queries about H2O.ai products.
  • Integrated Python APIs for H2O Wave and H2O GPT Enterprise, engineering a query controller that routes questions by product type.
  • Improved internal knowledge discovery and documentation accessibility across teams.

Teaching

I have served as a teaching assistant and course leader for Berkeley’s core computer science and data science curricula, with responsibilities spanning lecture development, infrastructure, and pedagogy design.

Term Course Role
Fall 2025 Data C88C — Computational Structures in Data Science Head uGSI (Lead TA)
Spring 2025 Data C88C — Computational Structures in Data Science uGSI (Teaching Assistant)
Fall 2024 Data C88C — Computational Structures in Data Science uGSI (Teaching Assistant)
Summer 2024 CS 61A — Structure and Interpretation of Computer Programs uGSI (Teaching Assistant)

Leadership & Service

Contact

Email: isabelle.ng@berkeley.edu

GitHub: isabe11eng · LinkedIn: /in/isabelle-ng-185875262/ · OpenReview: profile

© 2025 Isabelle Ng