Education

• York University, Ph.D. of Software Engineering
  Sep. 2020 – Now

• University of Waterloo, Master of Software Engineering
  Dec. 2017 – May. 2019

Publication

• CoCoFuzzing: Testing Neural Code Models with Coverage-Guided Fuzzing
  Moshi Wei, Yuchao Huang, Jinqiu Yang, Junjie Wang, and Song Wang
  IEEE Transactions on Reliability (TR’22)\

• API Recommendation for Machine Learning Libraries: How Far Are We?
  Moshi Wei, Yuchao Huang, Junjie Wang, Jiho Shin, Shiri harzevili Nima, and Song Wang
  ESEC/FSE 2022\

• CLEAR: Contrastive Learning for API Recommendation
  Moshi Wei, Shiri harzevili Nima, Yuchao Huang, Junjie Wang, and Song Wang
  ICSE 2022

• Automatic Unit Test Generation for Machine Learning Libraries: How Far Are We?
  Song Wang, Nishtha Shrestha, Abarna Kucheri Subburaman, Junjie Wang, Moshi Wei, and Nachiappan Nagappan
  ICSE 2021

• Yet Another Combination of IR-and Neural-based Comment Generation
  Yuchao Huang, Moshi Wei, Song Wang, Junjie Wang, Qing Wang
  IST 2022

• Coconut: combining context-aware neural translation models using ensemble for program repair
  Lutellier Thibaud, Pham Hung Viet, Pang Lawrence, Li Yitong, Wei Moshi, Tan Lin
  ISSTA 2020

• Survey on Automated API recommendation
  Wei, Moshi

Experience

Ph.D. Candidate, York University\

Sep. 2020 – Now

• Extracted and formatted data from Stackoverflow and the official Java documentation using BS4, followed by tokenization and EDA analysis using Scikit-learn, StanfordNLP, and NLTK, and visualisation using Seaborn and Pyplot.

• Created sourcecode pro-processing pipline with Python AST parser, APIs are extracted from a code snippet and tokenized with a custom tokenizer.

• Trained MLP models such as Seq2seq LSTM, BERT, CodeBERT, Pegasus, Text-CNN with Huggingface, Pytorch or Tensorflow with multiple loss functions for performance optimization such as contrastive learning, prototypical network, siamese network, and compared baseline models such as SVM, NB, KNN, and XGBoost build with sklearn based on multiple metrics such as MRR, MAP, precision, recall, and F1

Ericsson – Software Engineer (Intern)\

Sep. 2021 – Jan. 2022

• Joined the research team as a research intern for prototyping a web-based low-code webpage editing platform with web components, AngularJS, and GrapeJS.

Software Engineer – BI, Achievers\

May. 2019 – Sep. 2020

• Initiated and developed an awarded collaborative filtering-based user recommendation system with Spacy, Scikit-learn, and pandas. Run the system on 200 users for testing and feedback
• Developed and maintained daily ETL tasks with Centerprise ETL and PostgreSQL
• Speedup TB-level ETL data warehouse restoration time by 24 times by restructuring workflow using PostgreSQL and python