🚀 Joined Microsoft Research as a Research Intern
Started a research internship at Microsoft Research in Redmond, WA for Summer 2026.
☕️About Me
Yiyang “Diana” Wang is a Computer Science Ph.D. student at Georgia Tech, working in the Ka Moamoa Lab with Prof. Josiah Hester. Her research sits at the intersection of agentic AI, large language model (LLM) agents, and human–AI interaction, building intelligent systems that augment human capabilities and improve real-world decision making.
A central thread of her work is designing and evaluating multi-agent systems that are collaborative, reliable, and grounded in human needs. She has built LLM-agent frameworks to study complex social dynamics (AgentReview, EMNLP 2024) and relationship-centered, socio-collaborative AI companions for shared experiences (MASCOT, CompanionCast, and CultivAgents). She also advances the foundation-model stack itself — from robustness and prompt optimization for large language models (TextReg) to efficient self-distillation for LLMs (UniSD).
Complementing this, she develops multimodal sensing and machine learning systems for health and social good, translating raw signals into actionable insight — for example, fine-grained nicotine-intake estimation for smoking-cessation support (PuffEM, IEEE/ACM CHASE 2025). Across projects, she pairs rigorous evaluation and human-centered design with practical engineering (Python, ML pipelines, on-device systems) to ship research that drives product and real-world impact.
She is currently a Research Intern at Microsoft Research (2026), following a research internship at Dolby (2025) and software engineering internships at Amazon (×2) and HubSpot. She is actively seeking AI research / applied scientist internship and full-time opportunities, and is always glad to discuss new ideas and collaborations. ☕️
Interests
- Human-AI Interaction
- Large Language Models and LLM Agents
- Multi-Agent Systems
- AI for Health and Well-Being
Education
Ph.D. in Computer Science
Georgia Institute of Technology
Master of Human-Computer Interaction
Carnegie Mellon University
B.S. in Computer Science
Georgia Institute of Technology
Featured Research
Publications
(2026).
CultivAgents: Cultivating Relationship-Centered Multi-Agent Systems for Personalized Gardening.
arXiv.
(2026).
TextReg: Mitigating Prompt Distributional Overfitting via Regularized Text-Space Optimization.
arXiv.
(2025).
CompanionCast: Toward Social Collaboration with Multi-Agent Systems in Shared Experiences.
CHI 2026 Workshop.
(2025).
PuffEM: An E-cigarette Sleeve for Estimating User Nicotine Intake.
CHASE ‘25.
(2025).
Eliciting Design Guidelines of Paper-Based Tactile Interfaces for Eyes-Free Scenarios.
TEI 2025.
(2024).
AgentReview: Exploring Peer Review Dynamics with LLM Agents.
EMNLP 2024 (Oral).
(2024).
Scito2M: A 2 Million, 30-Year Cross-disciplinary Dataset for Temporal Scientometric Analysis.
Good Data @ AAAI 2025.
(2024).
Phantom Puffs: A Phantom Lung to Emulate Smoking Behavior.
OJRAT.
(2022).
Flexible Computational Photodetectors for Self-Powered Activity Sensing.
npj Flex. Electron..
(2020).
OptoSense: Towards Ubiquitous Self-Powered Ambient Light Sensing Surfaces.
IMWUT 2020.
Experience
Research Intern
Microsoft Research- Research Intern at Microsoft Research, Redmond, WA.
Ph.D. Research Intern
Dolby Laboratories- Developed CompanionCast, a multi-agent conversational AI framework designed to enhance social co-viewing experiences through LLM-driven companion agents.
- Integrated spatial audio processing into a multi-agent system to create immersive, realistic social interactions in group entertainment settings.
- Implemented an LLM-as-the-judge evaluation pipeline to iterate on agent alignment and conversational quality.
Research
Meta x CMU MHCI Capstone- Designed Nova, an LLM-based assistant, to enhance knowledge sharing for the advancement of data centers, overcoming challenges by applying UX research techniques and prototyping, and culminating in a collaborative presentation at Meta’s Fremont office.
Software Development Engineer Intern
Amazon- Delivered E2E pipeline using AWS (Lambda, State Machines) for natural language understanding models’ test set release that reduced manual effort and total testing time by 50%
Software Development Engineer Intern
Amazon- Optimized automated long-running end-to-end integration tests for internal model release pipeline to a single command on cloud terminal and shortened manual testing time by 8 hours
Software Engineer Co-op
HubSpot- Delivered Zoom call transcription service to more than 121,000 customers across more than 120 countries and improved conference archive and content accessibility
Software Engineer Intern
Cognosos- Enhanced customers’ portal experience in tracking over 100,000 assets across multiple industries at a start-up
Education
Ph.D. in Computer Science
Georgia Institute of TechnologySchool of Interactive Computing. Research in the Ka Moamoa Lab, advised by Prof. Josiah Hester.Master of Human-Computer Interaction
Carnegie Mellon UniversityGPA: 3.99/4.00. Advised by Prof. Sherry Tongshuang Wu.B.S. in Computer Science
Georgia Institute of TechnologyThreads: Intelligence and People. Cumulative GPA: 3.94/4.00, Major GPA: 4.00/4.00. Undergraduate research advised by Prof. Gregory Abowd and Prof. Jennifer Kim.
Awards
Mini-Seed Grant ($5,000)
College of Computing, Georgia Tech / Dolby ∙
September 2025
Co-wrote the proposal with my advisor; awarded $5,000.
NSF Student Travel Award
IEEE/ACM CHASE ∙
June 2025
1st Place, Graduate Poster Symposium (Junior Student category)
College of Computing, Georgia Institute of Technology ∙
May 2025
Selected Participant, Global Young Scientists Summit
National Research Foundation, Singapore ∙
January 2022
1 of 1,000 selected participants, nominated by Student Innovation & Undergraduate Research, Georgia Tech.
President’s Undergraduate Research Award (PURA) — Salary Award
Georgia Tech ∙
September 2021
1 of 80 undergraduate awardees in Fall 2021 to receive a stipend for conducting research.
Grace Hopper Celebration Student Registration Scholarship
AnitaB.org ∙
September 2020
Recent News
🌱 New preprint — CultivAgents
Released CultivAgents, a relationship-centered multi-agent system for personalized gardening.
📝 New preprint — TextReg
Co-authored TextReg, a method for mitigating prompt distributional overfitting.
📝 New preprint — UniSD
Co-authored UniSD, a unified self-distillation framework for large language models.
🎉 Passed my Ph.D. qualifier in CS — AI
Passed the Computer Science Ph.D. qualifying exam in the Artificial Intelligence area at Georgia Tech.
Visitors
