Resume

Sign in

Chemical Engineering Engineer

Location:
Kirkland, Washington, United States
Salary:
100K
Posted:
January 09, 2018

Contact this candidate

Zehua Wei (Wedward)

Seattle, WA 206-***-**** ac3yl7@r.postjobfree.com https://www.linkedin.com/in/wedward

Data Scientist with 6 years chemical engineering background, 5 years of coding experience, 2 years of machine learning engineering experience in consultancy and fin-tech start-up. Currently focus on Medical Image Processing, Natural Language Processing, and Deep Learning.

Technical Proficiencies

Image Segmentation

Natural Language Processing

Sentimental Analysis

TensorFlow

SpaCy

Word2Vec

OpenCV

Scikit-learn

Python

Keras

cNN

Pandas

Professional Experience

BIOPROBER CORP, Bellevue, WA, 09/2017 – Present

Data Scientist

Applied state of art deep learning technology to Medical AI. BioProber Corp is Seattle R&D center of SonoScape Medical Corp, developing next generation ultrasound system.

Key Accomplishments:

Trained a light-weighed convolutional neural network to extract obstetrical information (like head circumference) using TensorFlow, achieved real-time image processing (400*400 60 Hz video without GPU acceleration).

Model outperformed the U-Net image segmentation neural network (both have a dice score of 0.97), by using 80% less computational resource, and no manual labelled training mask data.

Testing different deploy method like Nvidia TensorRT 3.0 inference engine, TensorFlow Serving, also tested model’s performance on Nvidia Jetson TX2 embedding system.

LUXOFT USA, Kirkland, WA, 03/2017 – 07/2017

Data Scientist

Data Science Consultant in the Big Data team, Global Center of Expertise(CoE) Luxoft (NYSE: LXFT). Our team served clients like SoftBank(Automotive), CenturyLink(Telecom), CIBC(Financial Services), etc.

Key Accomplishments:

Prototyped CenturyLink’s field engineer chatbot project in an agile team in three months, developed a “query knowledge base” chatbot based on MS Bot Framework, Azure QnA maker and pdf splitter.

KEESUN TRADING LLC, Cambridge, MA, 09/2016 – 12/2016

NLP Software Development Engineer Intern

Built a sentimental analysis product in Keesun Trading’s natural language processing (NLP) team with top researchers from CMU and UW on GitLab. Applied Machine Learning technologies to evaluate the news articles’ sentiment score.

Key Accomplishments:

Designed a software package using classifiers built in Scikit-learn, on top of hand-engineered linguistic features, such as sentiment polarity lexicon features extracted by spaCy, an industrial-strength NLP library.

Developed a DL module with Keras, using GloVe (Wikipedia-trained word vectors) as the embedding layer.

Our linguistic model got 88.2% accuracy on IMDB (marked as our benchmark performance), DL models reached the benchmark after 2 epochs of training, and the ensemble DL model achieved 90.6% after 4 epochs.

Reported that cNN architecture in our application was leading to fast convergence and quick overfitting; Bi-direction LSTM architecture cost 10x training time, and it tended to be underfitting the training data.

Education

Master of Science – Chemical Engineering, 2016 University of Washington, Seattle

Concentration: Computational Chemistry

Bachelor of Science – Applied Chemistry, 2013 Beihang University, Beijing, China

PRIOR CAREER Experience

E-SCIENCE INSTITUTE, UW SEATTLE

Data Science Trainee

Engineered 10 features based on known slave trade voyage records, classified 251 slave trips out of 1651 unknown individual trips from 240k ship logs between 1750-1850, quantified 10x slave trips comparing to previous Kaggle study.

CHEMICAL ENGINEERING DEPARTMENT, UW SEATTLE

Research Assistant in Computational Chemistry

Built molecular models with Material Studio, analyzed the models with Gromacs-OpenMPI (MD simulation package) on UW Hyak HPC, developed clustering methods to find steady peptide GrBP5 conformations.

BEIHANG-YUANZI MOUNTAINEERING EXPEDITION

Chief Mountain Photographer

Followed and photographed the climbing team’s 7-day expedition to Mt. Qizi (6206m) in Tibet, after 3-month training.



Contact this candidate