Zhengyan Zhang

Boston, Massachusetts, United States

Join Prog.AI to see contacts

Summary

🤩

Rockstar

🎓

Top School

Zhengyan Zhang is a data-savvy software engineer with 10 years of experience bridging accounting/finance and applied machine learning, currently based in Boston. With a postgraduate background in business analysis and strong proficiency in R, Python, and Tableau, she translates financial and operational questions into robust data solutions. Her open-source contributions include work on high-profile NLP projects like TsinghuaAI's CPM and ERNIE—where she enhanced zero-shot classification, evaluation scripts, and result reporting—demonstrating a focus on model evaluation and practical deployment. She has also extended network-embedding tooling by adding GraRep and TADW to the OpenNE library, improving algorithm coverage and visualization examples. Comfortable across the stack, Zhengyan combines domain fluency in finance with hands-on ML engineering to deliver reproducible, evaluation-driven models.

10 years of coding experience

graduate student, graduate student at Northeastern University

Github Skills (19)

algorithm10

pytorch10

python10

evaluation10

machine-learning10

machine-learning-algorithms10

deep-learning10

natural-language-processing10

nlp10

scripting9

scikit9

bert9

scikit-learn9

trainings9

modeling9

Programming languages (2)

TeXPython

Github contributions (5)

thunlp/ERNIE

May 2019 - Jul 2022

Source code and dataset for ACL 2019 paper "ERNIE: Enhanced Language Representation with Informative Entities"

Role in this project:

Full-stack Developer

Contributions:40 commits, 3 PRs, 29 pushes in 3 years 2 months

Contributions summary:Zhengyan primarily modified the codebase related to the ERNIE model, focusing on evaluation scripts for various tasks. They updated parameter names and added evaluation functionalities, including a new evaluation script for FIGER. They also made changes to the output format of results and included a scoring script to assess the model performance. These modifications suggest a focus on improving and refining the model's evaluation and reporting capabilities.

nlprepresentationinformativedataseternie

TsinghuaAI/CPM-1-Generate

Nov 2020 - Jan 2023

Chinese Pre-Trained Language Models (CPM-LM) Version-I

Role in this project:

ML Engineer

Contributions:35 commits, 3 PRs, 32 pushes in 2 years 2 months

Contributions summary:Zhengyan primarily contributed to the zero-shot classification functionalities within the CPM-1-generate repository. Their work included modifying the `generate_samples.py` file to incorporate temperature adjustments in the softmax calculation. The user also added a new `zero-shot-cls.py` script and associated configuration files for performing zero-shot classification on OCNLI, IFLYTEK and TNEWS datasets, demonstrating a focus on model evaluation and application. Further contributions involved creating shell scripts to facilitate zero-shot classification testing with varied configurations.

nlpchinesepre-trained-language-modelsbertpre-trained

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