Huace (Josh) Yu

New Haven, Connecticut · joshua.yu@yale.edu

Print this page (Cmd + P) and choose “Save as PDF” for a clean copy.

Education

Yale University — New Haven, CT

Aug. 2026 – May 2030

Undergraduate at Yale College. Computer Science + Psychology.

Carnegie Mellon University — Pittsburgh, PA

Summer 2024

Summer Pre-College Program (Programming 15-112, Linear Algebra 21-241; A in both).

Beijing No. 4 High School, International Campus — Beijing, China

Sep. 2023 – May 2026

Awards: Principal's Award (Valedictorian).

Research and academic experience

Yale SOM — Research Assistant to Prof. Beth Anne Helgason, New Haven, CT

Oct. 2026 – Present
  • Developing methods to detect paltering in AI models, where a model avoids false statements but phrases its answers so the user is led to a false belief.

Multi-Paper Reasoning in LLMs — with Prof. Arman Cohan, New Haven, CT

Oct. 2026 – Present
  • Exploring how LLMs reason across multiple scientific papers.

Self-View Convergence in Human–AI Interaction — Project Lead, Beijing, China

2025 – 2026
  • Studied how AI agents influence human self-perception by building a multi-agent simulation in which LLM agents and human participants engaged in self-revelatory dialogue, finding measurable self-view convergence in social contexts but not perceptual ones.

CTB Innovation Competition — Medication Management Mini Program, Project Lead, Beijing, China

2023 – 2026
  • Led a team as project head to design and evaluate a mobile mini program for medication management in chronic kidney disease patients; paper accepted into the CTB database, project advanced to global finals and awarded third place.

Work experience

Developer — Yale Computer Society, New Haven, CT

Oct. 2026 – Present
  • Develop Yale Meals, an open-source app for Yale dining menus and meal statistics.
  • Used by a large share of students on campus.

Algorithm Research Intern — StepFun, Beijing, China

May 2026 – Aug. 2026
  • Refining a deepfake audio detection model within the company's data processing pipeline through pruning and retraining.
  • Conducting stress tests across services in the data processing pipeline and performed quality assessments to identify bottlenecks and weak points.
  • Overhauled data curation, centralized company audio data from spreadsheets into a web-based platform.

Skills and interests

Programming
Python (primary), Java, JavaScript, C++, Swift
Languages
English (native), Chinese (native)
Machine learning
Mostly self-taught via online courses and personal projects