Lichi Li

Sleepy almost surely!

I am an ML tech lead at Cisco, leading the design of numerous research and production agentic AI projects that help customers defend against software vulnerabilities. Previously I was a senior ML engineer and ML tech lead at Mercari, led and supported the development of numerous production and internal projects there. I also contributed to vehicle onboard UX ML and offboard causal inference at Mercedes-Benz Research & Development North America and automated efficiency analysis at BDP EnviroTech.

Formerly Bayesian in training and now deep learning in practice, I focus on science-driven applied machine learning research in building interpretable, powerful, robust and scalable machine learning algorithms. My expertise grew from extensive algorithmic design as well as the art in training dataset design with respect to each domain, application and use case.

Throughout my industrial service, I explored a diverse set of problem settings and methodologies, ranging from edge-bound to cloud-hosted models, from multimodal encoding to sequential recommendation and generative retrieval, from internal-facing user action behavioral modeling, automatic testing-data generation to external-facing online and offline production microservices and streaming or batch pipelines, be it deep, causal or probabilistic. I also actively work with academic researchers to push the state of the art understanding in large language models, recommendation systems as well as other related topics.

Previously advised by C. Lee Giles in information retrieval, then briefly and gratefully studied probabilistic machine learning from Tony Jebara, David M. Blei, Andrew Gelman, Stephan Mandt, and Laurent Charlin. I majored in Mathematics at Penn State University and subsequently Computer Science at Columbia University.

news

Sep 15, 2024	Joined Cisco! Starting to learn how vulnerabilities work together with AI agents.
Feb 25, 2020	Joined Mercari! Will begin to work more on deep learning on the cloud.
Aug 28, 2017	Joined Mercedes-Benz Research & Development North America! Looking forward to driving and working on a wide spectrum of cars.

selected publications

ArXiv

IndustryEQA: Pushing the Frontiers of Embodied Question Answering in Industrial Scenarios

Yifan Li, Yuhang Chen, Anh Dao, Lichi Li, Zhongyi Cai, Zhen Tan, Tianlong Chen, and Yu Kong

May 2025

arXiv Bib Code Dataset

@misc{li2025industryeqa,
  title = {IndustryEQA: Pushing the Frontiers of Embodied Question Answering in Industrial Scenarios},
  author = {Li, Yifan and Chen, Yuhang and Dao, Anh and Li, Lichi and Cai, Zhongyi and Tan, Zhen and Chen, Tianlong and Kong, Yu},
  year = {2025},
  month = may,
  eprint = {2505.20640},
  archiveprefix = {arXiv},
  primaryclass = {cs.CV},
  dataset = {https://huggingface.co/datasets/IndustryEQA/IndustryEQA},
}

ACL 2025

Towards Generating Controllable and Solvable Geometry Problem by Leveraging Symbolic Deduction Engine

Zhuoxuan Jiang, Tianyang Zhang, Peiyan Peng, Jing Chen, Yinong Xun, Haotian Zhang, Lichi Li, Yong Li, and Shaohua Zhang

Jun 2025

arXiv Bib Code

@misc{jiang2025generatingcontrollablesolvablegeometry,
  title = {Towards Generating Controllable and Solvable Geometry Problem by Leveraging Symbolic Deduction Engine},
  author = {Jiang, Zhuoxuan and Zhang, Tianyang and Peng, Peiyan and Chen, Jing and Xun, Yinong and Zhang, Haotian and Li, Lichi and Li, Yong and Zhang, Shaohua},
  year = {2025},
  month = jun,
  eprint = {2506.02565},
  archiveprefix = {arXiv},
  primaryclass = {cs.AI},
}

KDD 2025

MerRec: A Large-scale Multipurpose Mercari Dataset for Consumer-to-Consumer Recommendation Systems

Lichi Li, Zainul Abi Din, Zhen Tan, Sam London, Tianlong Chen, and Ajay Daptardar

Feb 2024

arXiv Bib Code Dataset

@misc{li2024merrec,
  title = {MerRec: A Large-scale Multipurpose Mercari Dataset for Consumer-to-Consumer Recommendation Systems},
  author = {Li, Lichi and Din, Zainul Abi and Tan, Zhen and London, Sam and Chen, Tianlong and Daptardar, Ajay},
  year = {2024},
  month = feb,
  dataset = {https://huggingface.co/datasets/mercari-us/merrec},
  eprint = {2402.14230},
  archiveprefix = {arXiv},
  primaryclass = {cs.IR},
}

ICWS 2014

A Web Service for Scholarly Big Data Information Extraction

Kyle Williams, Lichi Li, Madian Khabsa, Jian Wu, Patrick C. Shih, and C. Lee Giles

In 2014 IEEE International Conference on Web Services , 2014

Bib HTML Code

@inproceedings{williams2014,
  author = {Williams, Kyle and Li, Lichi and Khabsa, Madian and Wu, Jian and Shih, Patrick C. and Giles, C. Lee},
  booktitle = {2014 IEEE International Conference on Web Services},
  title = {A Web Service for Scholarly Big Data Information Extraction},
  year = {2014},
  month = {},
  pages = {105-112},
  keywords = {Data mining;Big data;Web servers;Information retrieval;Databases;Hamming distance;Web service;information extraction;scholarly big data;CiteSeerExtractor},
  doi = {10.1109/ICWS.2014.27},
}