Xinliang (Frederick) Zhang | 张心亮
Researcher at Google DeepMind, Gemini Team | CSE Ph.D.
Gradient Canopy
2000 N Shoreline Blvd
Mountain View, CA
Hi there, I am Xinliang (Frederick) Zhang. I am from Hangzhou. My Chinese name is 张心亮. Usually, I go by Frederick.
Currently, I am a Research Engineer at Google DeepMind in the Gemini team, working on Gemini Personalization.
I am a recent CSE Ph.D. graduate from University of Michigan (UM), where I was advised by Lu Wang. I was a member of LAUNCH Lab and affiliated with Michigan AI Lab. I received my B.S. degrees from The Ohio State University (OSU) in CSE and ISE, and a minor in Math. I also obtianed my B.Eng. degree from Sichuan University (SCU) in IE. I was fortunate to have Huan Sun as my advisor at OSU and proud of being part of SunLab. I was also privileged to work closely with Marie-Catherine de Marneffe.
I spent some fun time as a reserach scientist intern at Google DeepMind (Gemini/GenAI) (2025), Adobe Research (2024) and Bloomberg AI (2023).
My reserch interests lie in the fields of Natural Language Processing (NLP), Large Language Model (LLM) and Machine Learning (ML). I am also a regular reviewer for NLP venues.
- I am actively investigating
LLM Personalization(via memory mechanisms),System-2 ThinkingandEfficient LLMs. I am also interested inCode Models. - During the first half of my PhD tenure, I focused on
Structured LearningandMulti-granularity Information Extraction, as well as Narrative Understanding. - I used to work on Natural Language Inference, Information Retrieval, and Question Answering & Generation as an undergraduate researcher.
My work received Best Paper Award at BIBM 2021. I was awarded Outstanding Reviewer at EMNLP 2024 and NAACL 2022.
I am fortunate to have been mentoring so many talented young brains, and passionate about teaching!
- Yimin Wang (2025-2026), Incoming Ph.D. student @UCLA
- Yufan Wu (2024), Now Ph.D. student @OSU
- Xiaoyan Bai (2024), Now Ph.D. student @UChicago
- Jiayi Zhang (2021-2022), Now Ph.D. student @NEU
- Changyuan Qiu (2021-2022), Now Ph.D. student @UW
More about me:
- I am passionate about language learning, especially East Asian languages. I am almost always fascinated with the stunning beauty of languages.
- I am amazed by the Brushtalk in ancient times, and I hope the NLP advancements I am dedicted to will help realize a new form of Brushtalk across all nations in modern times. While pseudo-Chinese might be a starting point, what would be the next chapter to be more inclusive?
- Outside work, I enjoy J-Pop. I am a longtime fan of 乃木坂46/Nogizaka 46.
If you are interested in both AI and Idols/アイドル, let's connect!
I am looking for one/two self-motivated (PhD) student to work with on topics like LLM personalization and LLM thinking (GPUs are available for use). Please contact me directly, if interested. Warning: This is not an internship opportunity with GDM.
You may reach out to me at xlfzhang@{google.com,umich.edu}.
Alert: My homepage will be hosted at a different domain due to my recent graduation from UMich. The migration is expected to complete by the end of 2026 Summer. Stay tuned!
news (news archive)
| Apr 6, 2026 | Stoked to share that I have joined Google DeepMind as a research engineer. I will work on Gemini Personalization in the Gemini team. |
|---|---|
| Apr 5, 2026 | Our paper Do LLMs Really Need 10+ Thoughts for “find the time 1000 days after today?” Towards Structural Understanding of LLM Overthinking. got accepted to ACL 2026! Thanks to my mentors and collaborators at Google DeepMind. Paper is available online. |
| Apr 5, 2026 | Our paper Logit Arithmetic Elicits Long Reasoning Capabilities Without Training got accepted to Findings of ACL 2026! Congratulations to my collaborators. Paper and codes are available online. |
| Mar 18, 2026 | I have successfully defended my dissertation, titled Events, Discourse and Personalization: On Challenges and Methods in Computational Narrative Understanding! My dissertation will be available online during 2026 summer. Thanks to my committee members—Lu (Chair), Rada, David and Dallas—for their insights and support! |
| Oct 25, 2025 | Our paper FINDR: A Fast Influential Data Selector for NL2Code Pretraining got accepted to IJCNLP-AACL 2025! Paper is available online. |
| Oct 10, 2025 | Check out my recent pre-print paper in collaboration with Google DeepMind: Do LLMs Really Need 10+ Thoughts for “find the time 1000 days after today?” Towards Structural Understanding of LLM Overthinking. We conduct dual-axes benchmarking and structural understanding of LLM overthinking by introducing the first-ever sub-thoughts analyzer, TRACE. |
| Oct 10, 2025 | Check out our recent pre-print papers: Logit Arithmetic Elicits Long Reasoning Capabilities Without Training (we strive to elicit long thinking capability from a non-thinking LLM through light-weight logit arithmetics which requires no training), and LiveOIBench: Can Large Language Models Outperform Human Contestants in Informatics Olympiads? (we introduce LiveOIBench, a challenging IOI-level benchmark for assessing LLMs’ competitive programming abilities). |
| Sep 25, 2025 | Excited to give an invited talk at Google DeepMind, sharing my insights on LLM personalization. |
| Aug 25, 2025 | I am GSIing EECS 445 (Introduction to Machine Learning) this semester. Look forward to navigating the ins and outs of machine learning techniques with young brains. |
| Aug 20, 2025 | Our paper PRIME: Large Language Model Personalization with Cognitive Dual-Memory and Personalized Thought Process got accepted to EMNLP 2025! Thanks to my advisor. Paper and codes will be available online soon. |
| May 27, 2025 | Stoked to start my 3rd and find PhD internship at Google DeepMind. I will work with the Gemini/GenAI team. |
| May 19, 2025 | Check out my recent pre-print paper: PRIME: Large Language Model Personalization with Cognitive Memory and Thought Process! We propose a unified framework to tackle LLM personalization challenge, PRIME, built upon cognitive dual-memory model and augmented with novel personalized thinking. |
| Apr 21, 2025 | I have passe my dissertation proposal! My dissertation is titled Events, Discourse and Impact: On Challenges and Methods in Computational Narrative Understanding. Stay tuned for my final dissertation. |
| Mar 17, 2025 | I have officially formed my dissertation committee! Thanks to Rada, David and Dallas who agree to serve on my committee. Stay tuned for my dissertation title. |
| Feb 15, 2025 | Check out my recent pre-print paper: FINDR: A Fast Influential Data Selector for NL2Code Pretraining (We design a fast influential data selection strategy for LLM pretraining in NL2Code space, using code-aware features and informative priors-driven feature reweighting)! |
| Jan 30, 2025 | I have passed JLPT N4! It has been so many years since I started learning Gojūon (五十音) when I was just a middle school student. |
| Jan 8, 2025 | I am GSIing EECS 487 (Introduction to Natural Language Processing) this semester. Look forward to meeting next generation NLP talents. |
| Nov 14, 2024 | Awarded Outstanding Reviewer at EMNLP 2024! |
| Nov 12, 2024 | Attend EMNLP 2024 @Miami. I present my paper Narrative-of-Thought at the poster session! Feel free to check it out if you are looking a powerful prompting algorithm tailored to temporal reasoning. |
| Oct 10, 2024 | Selected as University-level Finalist and nominated for Two Sigma PhD Fellowship! |
| Sep 19, 2024 | Our paper Narrative-of-Thought: Improving Temporal Reasoning of Large Language Models via Recounted Narratives got accepted to Findings of EMNLP 2024! Thanks to my advisor. Paper and codes are available online. I can’t wait to see you guys in Miami this November! |
| Aug 26, 2024 | I am GSIing CSE 592 (Artificial Intelligence Foundations) this semester. It has been three years and a half since I TAed last time. |
| Jun 16, 2024 | Attend NAACL 2024 @Mexico City, Mexico, my first international in-person conference experience. I present my paper MOKA at the poster session! Feel free to check it out if aligning AI with human values and AGI are in your mind! Also, this Friday (06/21) also marks my fifth year in NLP research. Congrats to myself! |
| May 16, 2024 | Our paper ULTRA: Unleashing LLM’s Potential for Event Argument Extraction through Hierarchical Modeling and Pairwise Self-Refinement got accepted to Findings of ACL 2024! Thanks to my mentor and collaborators at Bloomberg. Paper will be available soon. |
| May 13, 2024 | Pleased to start my 2nd PhD internship at Adobe Research. I will work with Dr. Victor Bursztyn. |