Climate Risk Modeling Manager · Scotiabank

Astrid Zeng

Quantitative Researcher & Risk Modeler

MSc Statistics (Financial Modelling) Graduate at Western University, specializing in credit risk, climate risk frameworks, and machine learning applications in Quantitative finance. Bridging quantitative rigor with regulatory compliance — from IFRS 9 credit loss modeling to ESG risk assessment for institutional portfolios.

Personal Portfolio (Selected)
GOOGL
—
—
TSLA
—
—
MSFT
—
—
BTC LIVE
—
—
View Projects → Get In Touch
Mar 2026 · Incoming
Climate Risk Modeling, Manager
Scotiabank
May – Aug 2025
IT Risk and Strategy Intern
Aviva Canada
Sep 2024 – Oct 2025
MSc Statistics, Financial Modelling
Western University
Jun 2023 – Jan 2024
Research Analyst Intern
X-Institute
Jan – Jun 2023
ESG Intern
Cushman & Wakefield
01

Skills

Credit & Market Risk
Risk Modeling
PD / LGD / EADIFRS 9 / ECLBasel II / IIIStress TestingScenario AnalysisVaRFRTBAUC / KS
Climate & ESG
Sustainable Finance
Climate RiskESG QuantificationTCFD FrameworkOSFI ComplianceMSCI ESGCarbon MetricsTransition RiskPhysical Risk
Quantitative Methods
Statistical & Machine Learning
RegressionNeural NetworksMonte CarloTime SeriesBlack-ScholesRegularizationFeature EngineeringScorecard Modeling
Programming
Technical Tools
PythonC++RSQLpandas / NumPyscikit-learnTensorFlow / KerasExcel / VBAJira API
Data & Reporting
Analytics & BI
Power BIDashboard DesignAutomated ReportingData ValidationKPI MonitoringRisk IndicatorsStakeholder Comms
Domain Knowledge
Finance & Regulation
Derivatives PricingOptions GreeksFinancial ModelingP&L AttributionOSFI GuidelinesModel ValidationReal Estate Finance
02

Projects

01
Random ForestQuantitative Trading
Algorithmic Trend Prediction via Multi-Factor Random Forest
Developed an end-to-end pipeline to predict 5-day forward price direction. Engineered 20+ alpha features including momentum, mean-reversion (RSI/Stochastic), and volume-based indicators (OBV).
  • Implemented TimeSeriesSplit cross-validation to eliminate data leakage and ensure chronological integrity
  • Optimized Random Forest hyperparameters and utilized class_weight='balanced' to mitigate label imbalance
  • Built a live dashboard for real-time inference, outputting standardized directional probabilities $P(up/down)$
20+
Indicators
5yr
OHLCV Data
02
Credit RiskIFRS 9
Credit Risk Modeling & Expected Credit Loss Estimation
End-to-end credit risk pipeline estimating PD, LGD, and EAD in compliance with Basel II/III and IFRS 9. Processed 373,028 loan records with feature engineering and logistic regression scorecard modeling.
  • Stress testing and model validation using AUC and KS metrics
  • 40% improvement in Python pipeline runtime via code optimization
  • Model validation report with regulatory risk-focused insights
373K
Loan Records
40%
Runtime ↓
03
VolatilityNeural Network
Improving Implied Volatility Prediction through Neural Networks
Hybrid approach integrating Black-Scholes financial theory with an MLP neural network. Processed 12,528 options samples incorporating Delta, Vega, and Gamma as features to overcome B-S limitations.
  • 47% reduction in IVRMSE vs Black-Scholes baseline
  • Dropout, regularization, and early stopping for robustness
  • Applications in trading book risk monitoring and P&L attribution
47%
Error ↓
12.5K
Options Data
Get In Touch

Let's Connect

Open to discussions on investment, quant finance, and research collaborations.

lzeng72@uwo.ca