Xueguang 马雪光

Waterloo, Ontario, Canada

Join Prog.AI to see contacts

Summary

🤩

Rockstar

🎓

Top School

Xueguang 马雪光 is a PhD student in Computer Science at the University of Waterloo with eight years of engineering and research experience spanning industry internships at Meta, Microsoft, and Amazon and prior roles in software engineering and NFV testing. He focuses on information retrieval and dense retrieval methods, contributing backend work to prominent open-source IR toolkits Anserini and Pyserini—implementing WARC handling, dense TCT-ColBERT searchers, FAISS integration, and hybrid search support. His work bridges reproducible IR research and production-ready systems, combining deep algorithmic understanding with practical engineering (tests, refactors, and resource retrieval). Not obvious from his CV: he has repeatedly moved core retrieval functionality from prototype into robust library code, improving data handling and test coverage in widely used research toolkits. Based in Waterloo, he pairs ongoing PhD research with real-world ML/IR experience gained during top-tier research internships.

8 years of coding experience

2 years of employment as a software developer

The Experimental High School Attached to Beijing Normal University

Doctor of Philosophy - PhD, Computer Science, Doctor of Philosophy - PhD, Computer Science at University of Waterloo

Github Skills (8)

lucene10

javas10

machine-learning10

information-retrieval10

faiss10

python10

java10

testing9

Programming languages (3)

JavaJupyter NotebookPython

Github contributions (5)

castorini/pyserini

Jan 2021 - Jan 2023

Pyserini is a Python toolkit for reproducible information retrieval research with sparse and dense representations.

Role in this project:

Back-end Developer

Contributions:301 reviews, 166 commits, 353 PRs in 2 years 1 month

Contributions summary:Xueguang implemented core functionalities related to dense retrieval, specifically integrating the TCT-ColBERT dense retrieval method into the Pyserini toolkit. This included the initial implementation of a searcher for dense representations utilizing FAISS, along with the ability to encode queries, and support for hybrid search involving both dense and sparse representations. The contributions involved modifications to the existing code base, adding new classes and methods for handling dense retrieval and creating a robust system for dense retrieval.

record-linkagepythonrepresentationssparsesimilarity

castorini/anserini

May 2020 - Nov 2022

Anserini is a Lucene toolkit for reproducible information retrieval research

Role in this project:

Back-end Developer

Contributions:195 reviews, 26 commits, 50 PRs in 2 years 6 months

Contributions summary:Xueguang's contributions center on improving the codebase related to WARC (Web ARChive) collections. This involves refactoring existing code, addressing issues such as missing characters in WET indexes, and integrating new features. Specifically, the user has added methods to read files as strings and implemented tests for relevance judgments, while also incorporating methods to get resources from the index for better data retrieval.

reproducibleretrievalluceneinformation-retrievaljava

Find and Hire Top DevelopersWe’ve analyzed the programming source code of over 60 million software developers on GitHub and scored them by 50,000 skills. Sign-up on Prog,AI to search for software developers.

Request Free Trial