Portrait of Chuong Nguyen

Chuong Nguyen

B.S. in Computer Science, 2026 · UC San Diego

I am a recent Computer Science graduate and a research assistant at the Halıcıoğlu Data Science Institute. I study how AI agents can move from answering questions to actively expanding their understanding of the world: how they represent what they know, recognize what remains unexplained, decide which connections are worth investigating, and validate which discoveries deserve to change what they believe. My work spans knowledge representation, evaluation of LLM-generated interpretation, and agentic deep research systems.

chuongn194@gmail.com chn021@ucsd.edu LinkedIn GitHub CV

Research

I build AI agents that move from answering questions to expanding understanding, aligned with human interpretation. My interests span knowledge discovery and representation, curiosity-driven and abductive investigation, validating LLM interpretations against human judgment, agent systems design, and resource-aware reinforcement learning.

Knowledge Representation

Turning unstructured evidence into structured, relational knowledge an agent can reason over, without fixing that structure so tightly that unexpected relationships become impossible to discover (LOGOS).

Agentic Research & Curiosity

Building tool-using agents that research questions and synthesize evidence (Episteme), and exploring how agents could decide which question to pursue next.

Technology & Society

Prior work on parental concerns in online gaming platforms, using large-scale social media analysis and user studies to inform safer, more transparent platform design (Roblox Safety, IEEE ISTAS 2024).

Publications

Co-first authors are marked with *. My name is in bold.

2026

X Pi*, Q Yang*, C Nguyen*, H Shen

Bridge Human Interpretation and Machine Representation With Explicit Specification For Qualitative Data Analysis In LLM Era

International Conference on Machine Learning (ICML), 2026 — Position Paper (Poster)

ICML · arXiv · PDF

T Smith*, C Nguyen*, Q Yang, O Bandopadhyay, Y Su, N Polikarpova, X Pi

QualAlign: Benchmarking Automated Qualitative Coding Against Human Schemas

Conference on Language Modeling (COLM), 2026

OpenReview · PDF

2025

X Pi*, Q Yang*, C Nguyen*

LOGOS: LLM-driven End-to-End Grounded Theory Development and Schema Induction for Qualitative Research

arXiv preprint arXiv:2509.24294, 2025

arXiv · PDF

2024

A Smithwick*, C Nguyen*, E Gorial*, N Tran*, AM Flores, INS Munyaka

“Parent seeking Roblox Safety Help”: Comparing Parental Roblox Concerns to Roblox Offerings

2024 IEEE International Symposium on Technology and Society (ISTAS), pp. 1–9, 2024

IEEE Xplore

Research Experience

Research Assistant · Prof. Zhiting Hu's Group, HDSI

Mentors: Xinyu (Frederick) Pi and Qiyue Gao (PhD students)

LOGOS

Co-developed and led implementation of an LLM-driven framework that turns unstructured evidence (interview transcripts, legal texts, agent trajectories) into hierarchical schemas, building semantic clustering, relation-aware graph reasoning, and iterative refinement. Reaches an 88.2% matching rate against expert-developed schemas.

QualAlign (COLM 2026)

Co-built a benchmark for automated qualitative coding and ran the baselines. Organized evaluation metrics into lexical, semantic, and model-based interpretive categories, showing that metrics agree within a category but diverge across them.

Position Paper (ICML 2026)

Co-defined a framework that crosses levels of meaning-making with levels of modeling, and mapped existing automated qualitative-analysis systems onto it.

Episteme (in progress)

Co-building the agent harness for a deep research system: tool use, evidence synthesis, and evaluation.

Research Assistant · Ujima Lab

Advisor: Prof. Imani Munyaka

Beauty Filter

Analyzed facial enhancements from three filter applications (Snatched, Blochi Gora, Baby Girl), investigating cultural biases in appearance modifications. Migrated the Facial Landmarks Analysis Before and After Filter Enhancement algorithm to a remote server for parallel processing at scale.

Smart Mirror

Implemented DeepFace and Retina for facial recognition and racial classification on a Raspberry Pi smart mirror, delivering real-time facial analysis for a public art installation addressing racial bias.

Roblox Safety

Co-authored “Parent seeking Roblox Safety Help” (IEEE ISTAS 2024); conducted topic modeling on 10,000+ Reddit posts and surveyed 100 participants on parental concerns.

Early Research Scholars Program Intern

“Parent seeking Roblox Safety Help”: Comparing Parental Roblox Concerns to Roblox Offerings

Conducted topic modeling on 10,000+ Reddit posts and surveyed 100 participants to identify four primary parental concerns on Roblox. Proposed actionable solutions published at IEEE ISTAS and presented at the ERSP symposium.

Project Experience

Course, club, and team-based projects outside of formal research appointments.

Software Developer · Zooseeker

Designed and implemented the database schema for a zoo navigation Android application. Revamped the Android UI and maintained project documentation for the development team.

ACM AI

Credit Card Fraud Detector

Experimented with KNN, Random Forest, Decision Tree, SVM, and Logistic Regression for credit card fraud detection. Built and presented a fraud detection web application at the ACM symposium.

Education

University of California, San Diego

B.S. in Computer Science

Degree conferred June 2026 · GPA 3.803

Skills

Awards & Certifications