About MeSkillsExperienceProjectsCertificationsEducation
Hello, I'm Wei Yang 杨炜.
Data Scientist at The AES
Corporation, leading the Renewable Generation Forecasting platform, supporting LNG
Optimization, Turbine Fault Detection, LLM-based resume evaluation, and EVSE planning. Recommendations from
peer and direct manager can be found
on my LinkedIn profile 📩.
A self-proclaimed photographer. I love capturing pictures of architecture,
landscapes, stars, and wildlife. These photos not only showcase the world but also reflect a
part of myself at that moment.
I've set foot on six continents, with Antarctica being the only one left to explore.
Interested in Random
Matrix Theory, Data Science and Machine Learning.
Taught various courses in Mathematics and Statistics while doing Ph.D at Binghamton University. My
notes for Linear Algebra and
Calculus 3, can be found at the following links : 🔖(LinAlg), 🔖(Cal3) and 🔖(Stat).
Graduated from Baruch College, with studies concentrated in Economics, Financial and
Actuarial
Mathematics.
Skilled in multiple scripting languages, including Python, R, Javascript and Julia.
On this website, I will be creating notes on various materials for myself and others to learn or review.
The AES Corporation - Digital Solutions and Innovation
Data Scientist May 2023 -- Present, Houston TX
Wind & Solar Generation Forecasting
Spearheaded development of the AES Farseer renewable generation forecasting system for 30+ AES wind
and solar sites across the Americas. Achieved 30.4% mean absolute error reduction compared to vendor
models, generating $12.2M in operational savings by improving forecast accuracy.
Researched the impact of extreme weather events such as icing on wind generation, calibrated models
to the local weather patterns, and achieved a 45% reduction in forecast error under extreme weathers.
Enhanced model interpretability and strengthened stakeholder trust through the use of explainable AI
(XAI) techniques.
Solar Panel Dust Detection & Fault Detection System
Implemented EfficientNet architecture in PyTorch for solar panel dust detection, achieving a 16%
improvement over traditional CNN models.
Developed turbine fault detection models using Moving Window PCA, reducing fault detection time from
7 days to under 48 hours. The models reduced maintenance costs by over 45%.
LNG Optimization Model
Created stochastic dual dynamic programming (SDDP) models in Python/OR-Tools to optimize LNG
procurement strategy, increasing profit by 24% through dynamically balancing inventory, obligations,
and price volatility.
Other Projects
Collaborated with HR specialists on resume scoring models using Large Language Models (LLMs) &
Natural Language Processing (NLP) techniques. The model reduced the resume screening time of 300
applications from 2 weeks to 1 day.
Forecasting Commodity Trading
Renewable Energy
EVSE Locational Valuation LLMs NLP
Data Science Intern May 2022
- Aug. 2022
,
Indianapolis IN
Leveraged Random Matrix Theory and Modern Portfolio Theory to develop an internal package for
solving various portfolio
optimization problems.
Worked with teams of engineers and traders to leverage AI/ML for daily operations.
Portfolio Optimization Random Matrix Theory Modern Portfolio Theory Machine Learning
Binghamton University - Department of Mathematics and
StatisticsTeaching Assistance / Instructor 2016 - Present
Taught over 1000 students in Data Science, Probability, Statistics, Linear Algebra, and
single/multivariate Calculus
Created comprehensive study materials and interactive learning resources that improved student
comprehension of advanced mathematical concepts
Implemented data-driven assessment strategies, resulting in 20% average improvement in student
performance from initial to final exams
Provided individualized support through regular office hours and proactive student outreach,
maintaining consistently high student satisfaction
Instructor Binghamton
Graduate Student Organization at Binghamton Student Senator Sep. 2021 - Present
Attend senate meetings and committee meetings, on behalf of over 70 graduate students in department.
Attend Senate meetings and committee meetings, on the behalf of over 70 graduate students in the
department.
Communicate with colleagues in math department on pertinent issues raised during meetings.
Student Senator Graduate Student Organization
Queensborough Community College Math Tutor Feb. 2014 - Aug. 2016
Worked with students from disadvantaged backgrounds.
Guided students in developing study strategies and good study habits.
Led group and one-on-one tutoring session for struggling students.
Tutored students in courses including algebra, trigonometry, calculus and linear algebra.
Tutor Queensborough
Electricity Market Portfolio Optimization
Collected over 100 gigabytes of data from Independent System Operators (ISOs) through API and web
scraping.
Cleaned, transformed, and filtered data using results on random covariance matrices.
Generated recommendations for portfolios of long and short positions using filtered covariance
matrices. The results are
demonstrated to be more stable and outperforms unfiltered portfolios for over 93% of the time.
Portfolio Optimization
Handwriting Recognition Web Application
Balanced an unbalanced image data set by generating randomly perturbed images from the existing
ones,
the resulted data set contains over 2,500,000 28 × 28 pixel images with 62 labels.
Trained a Convolutional Neural Network model with the generated data set in Python with TensorFlow,
achieving 90% accuracy. The model is then exported as a JavaScript object, for web deployment.
Created a Handwriting Recognition Web Application with the trained model, and deployed to the
project
web page.
Handwriting Recognition Imbalanced Data TensorFlow Convolutional Neural Network Interactive Web App
S&P 500 Market Sectors
Reclassification
Performed Web Scraping on Wikipedia and Yahoo Finance to extract the data of the S&P 500 companies
over
the
last 10 years.
Cleaned, transformed the data, and denoised the transformed data using results from Random Matrix
Theory.
Performed hierarchical clustering on the denoised data.
The Reclassification provides a better understanding of the market, and enabled additional options for
portfolio optimization.
ClusteringRandom Matrix TheoryS&P 500
RandomMatrix.jl (a registered Julia package)
Developed a package to work with matrix-valued distributions.
Documented functionalities and deployed the documentation to the package documentation page
🔗.
Random MatrixPackage Documentation
Personal Website & Interactive Demos (This
website 😊)
September 2021 - Present
Designed a personal website from scratch using JavaScript and HTML, with the purpose of sharing my
projects and write about Mathematics and Data Science.
Created Interactive Demos to demonstrate ideas in Mathematics and Data Science.
This is a long term project, and I will be continuing to update the pages on this website.
Interactive Data Science Mathematics HTML JavaScript
Google Cloud Big Data and Machine Learning Fundamentals 🔗
EXAM FM Financial Mathematics, Society of Actuaries 🔗
Google Cloud ☁Python SQL Data Science Probability Financial Mathematics
Binghamton University, SUNY Ph.D. in Mathematics and Statistics (2023, ABD) Master of Science in Mathematics and Statistics (2018)
Dissertation Fellowship GPA : 3.91/ 4.00
Ph.D.Fellowship Mathematics StatisticsBinghamton
Baruch College, CUNY Bachelor of Art in Economics (2016) Bachelor of Art in Mathematics with a minor in Philosophy (2016)
Dean's List GPA: 3.77/4.00
Mathematics Economics Philosophy BaruchDean's List
Francis Lewis High School High School Diploma (2011)
Math Team