Wei Wang

I'm currently a Research Scientist in BCR CoreML ranking team at Meta. My work mainly focuses on creating new machine learning model for ad load personalization. My research interest include machine learning and its application in ads personalization experience, computational methodologies for efficient, accurate and highly scalable genomic analysis, especially for multiple sequence alignment(MSA), phylogenetic inference, and cancer biology.

News

  • [Jan 2021] Start a new position as Research Scientist at Meta, California, CA, USA
  • [Jun 2019] Oral presentation on IEEE BIBM 2019 Conference at San Diego, CA, USA
  • [Jun 2019] Student travel award of IEEE BIBM 2019 Conference at San Diego, CA, USA
  • [Jun 2019] Poster presentation on Evolution 2019 Conference at Providence, RI, USA
  • [Mar 2019] Poster presentation on Engineering Graduate Research Symposium at East Lansing, MI, USA [Won best poster prize]
  • [Oct 2018] Proceeding talk on RECOMB-CG 2018 at MAGOG-ORFORD (SHERBROOKE), QUEBEC, CANADA
  • [Oct 2018] Poster presentation on RECOMB-CG 2018 at MAGOG-ORFORD (SHERBROOKE), QUEBEC, CANADA
  • [Aug 2018] Poster presentation on BEACON conference at East Lansing, MI, USA
  • [Apr 2018] Poster presentation on Engineering Graduate Research Symposium at East Lansing, MI, USA
  • [Aug 2017] Poster presentation on BEACON conference at East Lansing, MI, USA
  • [Apr 2017] Poster presentation on Engineering Graduate Research Symposium at East Lansing, MI, USA [Won second poster prize]

Selected Publications

  • RAWR v1.0: a software suite for sequence-aware phylogenetic support estimation and other RAndom Walk Resampling tasks
    Julia Zhang, Wei Wang, Meijun Gao, and Kevin J. Liu
    Under review

  • The Impact of Multiple Sequence Alignment Error on Summary-based Phylogenetic Network Estimation
    Meijun Gao, Wei Wang, and Kevin J. Liu
    the 13th ACM International Conference on Bioinformatics, Computational Biology and Health Informatics, 2022 [Paper] [Source Code]

  • Build a better bootstrap and the RAWR shall beat a random path to your door: phylogenetic support estimation revisited.
    Wei Wang, Ahmad Hejasebazzi, Julia Zheng, Kevin J Liu
    ISMB/ECCB 2021 [Paper] [Source Code]

  • An Application of Random Walk Resampling to Phylogenetic HMM Inference and Learning
    Wei Wang, Qiqige Wuyun, and Kevin Liu
    IEEE transactions on nanobioscience [Paper] [Source Code]

  • An Application of Random Walk Resampling to Phylogenetic HMM Inference and Learning
    Wei Wang, Qiqige Wuyun, and Kevin Liu
    IEEE BIBM 2019 [Paper] [Source Code]

  • Non-parametric and semi-parametric support estimation using SEquential RESampling random walks on biomolecular sequences
    Wei Wang, Jack Smith, Hussein Hejase and Kevin Liu
    Algorithms for Molecular Biology. [Paper] [Source Code]

  • Non-parametric and semi-parametric support estimation using SEquential RESampling random walks on biomolecular sequences
    Wei Wang, Jack Smith, Hussein Hejase and Kevin Liu
    RECOMB-CG 2018. [Paper] [Source Code]

    • Experience

    [Jan 2022 - Now] BCR CoreML Ranking Team, Meta
     Research Scientist
     - Main contributor of the personalization modeling for ad supply release project, which is aiming at personalize the ad supply based user sensitivity to increase revenue gain with minimum engagement hurt.
    - Designed and developed the user-level sensitivity model to reduce the engagement hurt by 56% under the same revenue gain level.
    - Built entire workflow of user personalization model which has been generalized to multiple products, include ML problem definition, offline data collection, feature engineering, model iteration and improvement, online model performance validation and testing, model serving and maintenance.

    [Aug 2016 - Oct 2021] Department of Computer Science and Engineering, Michigan State University
     Research Assistant
     - Designed and developed SERES, a new sequential resampling approach for biological sequences, which has been successfully applied on support estimation of multiple sequence alignment and improved the accuracy of the estimated support by 10%.
    - Developed local genealogy inference approach with the SERES resampling approach.
    - Developed non-parametric sequential resampling approach RAWR for the support estimation of phylogenetic trees, which improved the PR-AUC performance by 35%.
    - Developed survival prediction model for triple-negative breast cancer based on gene expression data.
    - Developing reinforcement learning approach for support estimation of multiple sequence alignment.
    - Developing algorithm for phylogenetic inference using RNA-seq reads.
    - Developed desktop software, commend line tool and online service for phylogenetic support estimation utilizing SERES and RAWR resampling algorithms.

    [May 2019 - Aug 2019] LinkedIn, Sunnyvale, CA
     Summer Intern
     - Worked in the Anti-Abuse AI team for 3 months.
    - Developed name ranking model using character-based CNN for fake name detection.
    - The name ranking model reached 99.35% ROC AUC on 30 million test dataset.

    [Feb 2015 - Sep 2015] Department of Biomedical Engineering, University of Texas at Austin
     Research Assistant
     - Developed and refined preprocessing of large RNA-Seq data using clustering. Reduced sequencing error rate to 0.003%.
    - Built up human donor database to manage basic information and sequencing data of human donors.

    [Sep 2013 - Jun 2016] Key Laboratory of RNA Biology, Institute of Biophysics, Chinese Academy of Science
     Research Assistant
     - Built up classification model for mutation pattern of antibody repertoire of HIV infected patients based on large RNA-Seq data.
    - Identified patterns involved in different HIV disease progression from high-dimension gene expression data by clustering methods.