Machine Learning Data Analyst

Siyuan Nie

# ********@******.***.*** 412-***-**** ð brentluy-nie

Education

M.S., Carnegie Mellon University, Tepper School of Business Business Analytics (2024.8-2025.5) B.S., Xi’an Jiaotong-Liverpool University Mathematics and Finance (2020.9-2024.6) Skills

AI/ML: NLP, Computer Vision, Regression Analysis, Metric Development, Time Series Model Tools: Pandas, Bloomberg, Wind, Gurobi, Kafka, Tableau, NumPy, TensorFlow, A/B Test Programming: Python, SQL (MySQL), R and R Studio, MATLAB, LaTeX, PySpark Finance: Financial Ratio Analysis, Credit Risk Analysis, Merger Modeling Languages: English, Mandarin, and Chinese

Certificates: CFA-ESG, FRM-L1, TensorFlow Developer, Bloomberg BMC, Bloomberg ESG Experience

Machine Learning Engineer, Data Science

Westinghouse

Pittsburgh, PA

Jan 2025 – May 2025

Sensor Data Denoising: Addressed noise of reactor sensors data using wavelet transforms (PyWavelets) and 3 sigma outlier filtering, imputed missing values via MICE, enabling reliable ML inputs for production simulation.

Production Optimization: Implemented Bayesian-tuned XGBoost model with SHAP analysis, identifying fuel rod interval as key factor (34%impact), enabling 12.3% UPH(unit-per-hour throughput) lift via digital A/B testing.

Financial Modeling: Built defense-sector BOO model with tailored cost/revenue assumptions from DoD infras- tructure and reactor data; identified scale-dependent viability with IRR rising from –1.6% to 6.5% (10 units).

Skills & Tools: MICE, XGBoost, Regression Analysis, A/B Test, SHAP analysis, Python, Excel Data Scientist, Energy & Risk Management

Infaith, Internship

Beijing, China

Feb 2024 - May 2024

Coal Quality Recognition: Designed MobileNetV2-based classifier across 50k images using CLAHE and OpenCV, deployed Flask API with latency <10s latency, reducing manual inspection errors from 15.4% to 3.9%.

Liquidity Risk Scoring: Built PCA-based score on 15 SAP metrics across 21 subsidiaries, improving crisis detection (AUC 0.82), enabling dynamic buffers (risk score 10%), cutting incidents by 23.6% and freeing $8M.

Smart Mining Optimization: Built LSTM-based CO2 prediction system with infrared sensors, reducing crusher idle time by 18% and energy costs 3.5%, later scaled to 3 sites through Tableau-powered dashboards.

Skills & Tools: Transfer Learning, PCA, LSTM, CLAHE, OpenCV, Kafka, PySpark, SQL, Tableau Data Analyst, Finance and Risk

PwC, Internship

Changsha, China

Aug 2023 – Dec 2023

Data Analysis: Queried 126K records via SQL to analyze 3-year data across 5 parks; identified 2 zones with 35% lower yield/acre and recommended targeting high-value manufacturing, boosting new project output by 24.6%.

Data extraction: Automated ESG metrics collection via Python Web Crawlers, resolving 47 field inconsistencies through fuzzy string matching, delivering Excel-ready firm/individual data for evaluation.

Insurance Risk Modeling: Analyzed 100K policies with 26 factors via XGBoost (AUC 0.92), using SHAP explanations, removing 5 biased attributes while maintaining 95% F1-score to secure $1M deal.

Skills & Tools: SQL, XGBoost, SHAP analysis, DiCE analysis, Tableau, Financial Modeling, Excel Data Analyst, Finance and Risk

Green Light-Year, Internship

Shanghai, China

Jan 2023 - May 2023

Rating System: Designed 173-factor template using Analytic Hierarchy Process, integrating audited data to align Basel III with Chinese policies, achieving 15.3% accuracy gain in rating prediction across policy institutions.

Cross-Border Score Optimization: Identified 21 standardization factors from 20-year data(2Kfirms, 1.6Mpoints) via LASSO regression, elevating Chinese enterprises’ benchmark scores by 30% through structural bias reduction.

Skills & Tools: Analytic Hierarchy Process, Excel, LASSO Regression Data Analyst, Investment Bank

Soochow Securities, Internship

Suzhou, China

May 2022 – Sep 2022

Performed in-depth sensitivity analysis on historical price fluctuations of 17 raw materials, including lithium and aluminum, using data from Wind and Bloomberg to address exchange inquiry questions.

Processed and cleaned 5 years of financial data for 11 companies, verifying details of 1,581 directors and supervisors and 279 related parties. Conducted bidirectional checks of chronological ledgers and prepared issuer documentation.

Skills & Tools: Bloomberg, Due Diligence, Wind, Tableau Prep, SQL, Excel

Contact this candidate