SequenceO1
End-to-End Ultra-Long (100K) Sequence Modeling in Recommendation with Low-Rank Caching
Jia-Qi Yang 杨嘉祺
Algorithm Engineer · Researcher · Ph.D.
I am an engineer and researcher, with my primary research interests in machine learning and data mining. My work focuses on recommender systems, uncertainty quantification, and AI for science.
Featured Research
Semantic-Aware Grouping
Calibrating Neural Networks with Semantic Aware Grouping
Programmable Metasurfaces
Physics-informed Inverse Design of Multi-bit Metasurfaces
Publications
* equal contribution
SequenceO1: End-to-End Ultra-Long (100K) Sequence Modeling in Recommendation with Low-Rank Caching
Make It Long, Keep It Fast: End-to-End 10k-Sequence Modeling at Billion Scale on Douyin
Improving LLMs for Recommendation with Out-Of-Vocabulary Tokens
JobFormer: Skill-Aware Job Recommendation with Semantic-Enhanced Transformer
Physics-informed Inverse Design of Multi-bit Programmable Metasurfaces
COURIER: Contrastive User Intention Reconstruction for Large-Scale Visual Recommendation
RE-SORT: Removing Spurious Correlation in Multilevel Interaction for CTR Prediction
Beyond Probability Partitions: Calibrating Neural Networks with Semantic Aware Grouping
IDToolkit: A Toolkit for Benchmarking and Developing Inverse Design Algorithms in Nanophotonics
Normalizing Flows for Efficient Inverse Design of Thermophotovoltaic Emitters
Generalized Delayed Feedback Model with Post-Click Information in Recommender Systems
RID-Noise: Towards Robust Inverse Design under Noisy Environments
Capturing Delayed Feedback in Conversion Rate Prediction via Elapsed-Time Sampling
Corporate Relative Valuation using Heterogeneous Multi-Modal Graph Neural Network
Bottom-Up and Top-Down Graph Pooling
Deep Multiple Instance Selection
Towards Understanding Transfer Learning Algorithms Using Meta Transfer Features
Work Experience
Education
Visiting Student
2024.3 - 2024.12Imperfect Information Learning Team, RIKEN-AIP.
Supervisor: Prof. Masashi Sugiyama.
Doctor of Philosophy (Computer Science and Technology)
2021.9 - 2025.6LAMDA Group, Nanjing University.
Supervisor: Prof. De-Chuan Zhan.
Master of Engineering (Computer Science and Technology)
2018.9 - 2021.6LAMDA Group, Nanjing University.
Supervisor: Prof. De-Chuan Zhan.
Bachelor of Engineering (Electrical Engineering and its Automation)
2014.9 - 2018.6Talks
Delayed Feedback in Recommender Systems
Awards
Academic Recognition
- Outstanding Doctoral Dissertation Award of the School of Computer Science, Nanjing University, 2025
- KDD Outstanding Reviewer (top 10%), 2025
- Scholar/Travel Award, NeurIPS'22, AAAI'22, KDD'23, NeurIPS'23.
Scholarships & Research Programs
- Nanjing University Doctoral Students’ Innovation Capacity Leadership Program – Excellence Creation Project, 2024
- Postgraduate Research & Practice Innovation Program of Jiangsu Province, 2024
- Scholarship of China Scholarship Council (Visiting student at RIKEN-AIP), 2024
- Huawei Scholarship, 2023
- National Scholarship for PhD student, 2022
- Award for Outstanding Contribution, Huawei-Nanjing University Joint AI Laboratory, 2021
Competitions
- Champion(1st/2116), Alibaba Global Scheduling Algorithm Competition, 2018
- Most Exquisite Code Award, Huawei Code Craft, 2018
- KENSHO prize, HackHarvard, 2018
- Gold Award, Fastest solver of problem A, ACM/ICPC Asian Regional, HongKong Onsite, 2016
- Silver Award, ACM/ICPC Asian Regional, East Continent Final, 2016
- Gold Award, China Collegiate Programming Contest(CCPC), South China Invitational Contest, 2016
- Silver Award, China Collegiate Programming Contest(CCPC), Changchun Onsite, 2016
- Bronze Award, ACM/ICPC Asian Regional, Shanghai Onsite, 2015
Academic Service
Conference / Journal Reviewer
Conferences
Journals
Teaching Assistant
2023
Intelligent system design and application. with Prof. De-Chuan Zhan.
2021
Digital Signal Processing. with Prof. Han-Jia Ye.
2018
Advanced C++ Programming. with Prof. Shu-Jian Huang and Prof. Zhi-Qiang Zuo.