Yang Yang
Staff AI Research Scientist, Meta
- st2yang [at] gmail.com
- Google Scholar
I work on learning systems that improve with scale and feedback: foundation models for recommendation, RL post-training for LLMs, and AI agents for ML engineering. My work is deployed in recommendation and ads systems at Meta’s scale.
My PhD research at the University of Minnesota was in robot learning and won the ICRA 2022 Outstanding Student Paper Award.
Selected publications
Full list on Google Scholar
Recommendation systems
-
KDD 2026 Oral
Unity: Fully Self-Supervised Pretraining with Transformers for Recommendation
Masked pretraining on latent tokens that unify heterogeneous engagement features.
-
Preprint 2026
The Case Against Generation for Retrieval: Discriminative Language Models as Effective Retrievers
Two-tower retrieval with a shared LLM encoder, distilled from a cross-encoder teacher.
-
RecSys 2025
Enhancing Embedding Representation Stability in Recommendation Systems with Semantic ID
Content-based prefix-ngram Semantic IDs that replace random hashing.
-
KDD 2023
Breaking the Curse of Quality Saturation with User-Centric Ranking
Ranking with users as tokens instead of an unbounded item vocabulary.
AI agents
-
ICLR 2026
STARK: Strategic Team of Agents for Refining Kernels
LLM agents that search GPU kernel designs, guided by profiling feedback.
-
ICLR 2026 MAL-GAI Workshop Oral
ArchPilot: A Proxy-Guided Multi-Agent Approach for Machine Learning Engineering
Tree search that ranks candidate architectures by scores from proxy training runs.
Robot learning
-
T-RO 2024
Attribute-Based Robotic Grasping with Data-Efficient Adaptation
Attribute grounding learned in simulation, adapted with unlabeled images or one grasp.
-
ICRA 2022 Outstanding Student Paper Award
Interactive Robotic Grasping with Attribute-Guided Disambiguation
Attribute or pointing questions chosen by a POMDP to resolve ambiguous commands.
-
RA-L 2020 Code
A Deep Learning Approach to Grasping the Invisible
Pushing and grasping to retrieve an initially hidden object in clutter.
Experience
Staff AI Research Scientist
2022 to present
Meta
Research Intern
2020
Mitsubishi Electric Research Laboratories
Software Engineering Intern
2018
Google
Education
PhD, Computer Science
2022
University of Minnesota
BEng, Energy and Power Engineering
2015
Huazhong University of Science and Technology