Wesley Wei Qian

PhD Student in Machine Learning & Bioinformatics

<-- About Section -->

About Me.


my name is wesley

i study machine learning and bioinformatics

at university of illinois at urbana–champaign with prof. jian peng

and i enjoy photography ,   golf   and yoga 禪   quite a bit



Graduate Research Assistant

  • Work with Prof. Jian Peng on various tasks in machine learning and bioinformatics like secondary structure prediction, protein alignment, and domain fold prediction

  • Built and validate deep learning model like ConvNet, LSTM, ByteNet, and Auto Encoder, in Tensorflow and PyTorch



Software Engineering Intern

  • Summer 2016: created and designed an internal tool for mobile developers to investigate UI test failures in full speed by aggregating and synchronizing logs and videos with React.js front-end and Python/Go back-end

  • Summer 2017: built model to estimate the trip states (e.g. parking, waiting etc) for Uber Eats couriers, and developed a Kernel Conditional Random Field library drawing interest from multiple teams and won the first prize for Uber’s first internal machine learning poster session



Research Assistant

  • Conducted interdisciplinary research in computer science, biology and natural language processing with Prof. Pengyu Hong

  • Develop statistical machine learning model and engineer neural network to solve some challenging problems in protein structure understanding in biology and discourse parsing in linguistic

  • Selected as Jerome A. Schiff Undergraduate Fellows for my thesis



Google CodeU Participant

  • Participated CodeU, an exclusive dev program for high potential students to strengthen their skills

  • Built and presented the application at Google Tech Corner and won the runner-up for Engineer’s Choice



Teaching Assistant

  • The role has differed from course to course, but duties consistently entail office hours, grading, test grading, holding recitations, and review sessions

  • The courses are: Intro/Advanced Programing in Java, Data Structure, Structure and Interpretation of Computer Programs, Operating System, and Data Management System



Restaurant Revenue Prediction

  • Rank top 2% among 2257 teams from all over the world

  • Build multiple models using Ensemble, Linear Regression, Support Vector Machine, Regression Tree and Random Forest, run feature selection using Boruta, and engineer data using techniques involving data imputation and data visualization.




  • Jeeves is a voice-powered virtual assistant for everyday routines and the Runner-Up of AVIOS Mobile Speech Application Contest 2015

  • Features include: Reading email, current news, and weather, natural language understanding, dialogue management



Pattern Learning
and Protein Modeling

  • Designed an efficient graph matching algorithm, and used EM algorithm to build a probablistic mixture model to represent the learned patterns from graph

  • Project is supported by Jerome A. Schiff Fellowship for its application in protein 3d structure modeling



Discourse Parsing in Chineses

  • Crawl and preprocess social network text to develop various neural network for sentence vector representation or sentence relation classification

  • Customize word embedding solution for Chinese language features like its rich morphologies, and improve word embedding with better semantic representation using dictionary

  • Project is supported by Dean of Art and Science: Student/Faculty Collaborative Research Grant



DT in the House

  • Make Donal Trump speaks again

  • Build a stack LSTM RNN that speaks like Donal Trump by learning his speech one character at a time

  • Develop with Keras and Tensorflow backend



Contact Transfer

  • Created a contact transfer Android Application that transfers users’ contact and social platform info in a quick manner using NFC or QR Code

  • Worked with Android Beam API for NFC support, ZXing Project for QR code support and Parse backend in Android Studio

  • Presented the application at Google Tech Corner and won the runner-up for Engineer’s Choice



For more please visit