Nov 2023 – Aug 2025 · 1 yr 9 mos · Denver, Colorado, United States

• 1. Worked on the valuation of interest rate swap (fixed and floating leg). The valuation engine contains variety of interest rates (SOFR, US TSY), different day count conventions (ACT/365, ACT/360, 30/360) and different compounding frequency (continuous, daily, weekly, monthly, semi annual and yearly). Comparing it with CME Curve, Calypso & Polypath Curve.
• 2. In Calypso and PolyPaths,validated SOFR interest rate curve interpolation techniques, including linear interpolation, cubic spline, and monotone convex, to analyze their effects on interest rate swap pricing. Each method impacts the curve’s smoothness and stability differently, influencing swap valuations. Comparing these results helped assess which technique offers the most accurate pricing for interest rate derivatives.
• 3. Automate the sensitivity Analysis of MBS Portfolio (Polypath + ADCO Model) using Batch Calc process. The sensitivity parameter includes FICO Score, OLTV, OFICO, Orig Loan Size, Refinance Interest Rate, Non Refi rate and HPI
• 4. Optimize Generalized Hull-White Trinomial Lattice Model for bonds will call provision (callable bond) and BGM (Brace-Gatarek-Musiela) 3 Factor Model for mortgage, non structured product and non callable bonds. Wrote Python code to calibrate the model using Swaptions & Caps and minimized the error between market and model price.
• 5. Compared and analyzed different volatility models, including SABR and CEV, by evaluating their ability to capture implied volatility skew, term structure, and market dynamics. Assessed model performance based on calibration accuracy, stability, and suitability for interest rate derivatives (swaps and caps) asset class

Jun 2022 – Oct 2023 · 1 yr 4 mos · United States

• Project 1: CCAR/DFAST 2022 and 2023
• Automated, re-calibrated and re-developed the PPNR Models (C&I, CRE, NIB, IBC, Mortgage Refinance, Mortgage Purchase, Consumer Savings) for CCAR/D-FAST run. Performed sensitivity, scenario analysis, ongoing monitoring, back testing, quarterly forecasts, ad-hoc analysis on Region’s Bank Treasury’s portfolios. Proposed model changes as necessary such as overlays and performed macro-economic driver analyses for the CCAR/D-Fast 2022.
• Monitored model performance and communicated results to internal stakeholders (Model Risk Management & Validation, ALM, Investment Portfolio, and Econometrics Team).
• Project 2: Back testing of the Mortgage Loan Dynamics Model (LDM).
• Carried out back testing of the Agency (Fannie Mae, Freddie Mac, Ginnie Mae) and Non Agency AD&Co. LDM Model to understand the prepayment behavior of mortgage borrowers in QRM Enterprise Risk Framework.
• Project 3: Forecast Regions Treasury Deposit Products (Consumer Checking, Money Market, Savings, CDs)~$40B
• Cleaned, processed and filtered the data into a suitable format for analysis. Performed data modeling to identify trends, seasonality, and patterns in the data.
• Carried out feature selection technique by using forward selection, backward elimination and regularization techniques (ridge, lasso). Performed in sample and out of sample test and on approval, submitted the models to Model Validation Group for review.
• Project 4: Model for Predicting Day on Day Balance for Commercial Banking Group (approx. $36B Portfolio):
• ETL the data from Qlik software and combined the commercial clients from different industries.
• Applied K-means clustering algorithm to group commercial clients (~46k) in the form of clusters (k=3)
• Tried and tested various regression models (linear, ridge, lasso, elastic net, weighted least square) and ended up using weighted least square regression model for each cluster.

Jan 2021 – May 2022 · 1 yr 4 mos · United States

• During my assistantship, I worked on various projects. Few of them are listed below:
• 1) Valuation of Treasury Inflation-Protected Security (TIPS): Forecasted the inflation index (CPI) using ARIMA, SARIMA models and Constant Growth Model. Developed CIR & Vasicek interest rate model to generate yield curve and calculate the bond price by discounting the cashflows. Built challenger model to validate assumptions, conceptual soundness, formulas, and performance.
• 2) VaR and C-VaR Model and back-tested it: Constructed a portfolio of 10 banking stocks and calculated 1-day VaR at 99% and 95% CI (Variance-Covariance & Historical Simulation) and 1 day C-VaR at 99% and 95% CI to determine the expected portfolio loss. Back tested the model to evaluate the accuracy of the VaR model. Checked VaR breaches and documented the risk exposure.
• 3) Derivative Pricing Using Monte Carlo: Deduce an equation to price Auto Callable Barrier Reverse Convertibles which takes into account underlying asset, autocall features, barrier levels, redemption conditions etc., Used Geometric Brownian Motion (GBM) to simulate a large number of future price paths for the underlying asset. Monitor the simulated asset price paths to check if any barrier levels are breached. Discounted the cash flows using an appropriate risk-free interest rate and summed the value to get the estimated price of the ABRC.
• 4) Principal Component Analysis (PCA) on Interest Rates: Collect historical interest rate data from Bloomberg for set of maturity points along the yield curve. Apply PCA to identify the principal components. Assess each principal component in explaining the yield curve dynamics and interest rate movements.

Jun 2018 – Dec 2020 · 2 yrs 6 mos

• 1) Benchmark ML Model to calculate Probability of Default (PD) for Mortgage Borrowers
• Collected and processed 300K of mortgage data using Python (numpy, pandas, matplotlib, scikit-learn). This includes cleaning, validating and transforming data on borrower characteristics, loan performance, and macroeconomic factors for modeling.
• Carried feature engineering techniques (min-max scaler, one hot encoding, fine classing etc.,), feature selection (weight of evidence, information value, chi-square test etc.,) and narrowed 175 variables. Selected ML Algorithms (logistic, support vector machine, decision tree, & random forest) and performed on going monitoring (precision, accuracy, recall, f1 score, roc curve) for PD model.
• 2) Primary Secondary Spread (PSS) Model for Mortgages
• Built linear model for PSS using short-term and long-term treasury yields and incorporated volatility term (GARCH). Performed in sample and out of sample testing and benchmarking to evaluate model's accuracy and communicated the results with the ALM Team.
• 3) Model Validation for CCAR Models
• Analyzed data, including reviewing data quality, identifying data anomalies, and verifying data accuracy, reliability and completeness. Review the model documentation, including the model assumptions, methodology, limitations, and validate results.
• Conduct various validation tests to ensure compliance with regulatory requirements (FRB SR 11-7, SR 15-19, OCC 2011-12). Prepare reports summarizing the model validation findings, and recommend corrective actions and communicated with the stakeholders.

Jan 2018 – May 2018 · 4 mos

Dec 2016 – Dec 2016 · 0 mo · Rajasthan, India