EXPERIENCE

Mortgage Loan Default Model

PROJECT

June 2023 – Aug 2023

● Utilized Boston HMDA dataset to construct a Mortage Loan Defult Model employing Random Forest.

● Implemented Shapely Additive explanations (SHAP) to inteprate the model’s predictions and the key factors influencing defults.

● Presented results using SHAP summary plots and force plots to illustrate the impact of different features on global and individual predictions.

● SHAP analysis-driven recommendations led to a 3% decrease in the defult rate through targeted retention strategies.

Build RAG Agent with LLM Integration

PROJECT

Dec 2023 – March 2024

• Developed a Retrieval-Augmented Generation (RAG) agent by integrating Language Model

(LLM) capabilities, enabiling predictable user interaction and utilizing internal and external reasoning components for enhanced performance.

• Designed and implemented a robust dialog management and document reasoning system, ensuring efficient information structuring and state maintenance to facilitate smooth communication and interaction.

• Leveraged embedding models to enable effective similarity queries for content retrieval and dialog management, enhancing the overall functionality and usability of the system. Denoising and Forecasting Model Development

PROJECT

Aug 2023 – Dec 2023

• Developed a sophisticated model utilizing Fast Fourier Transformation (FFT) to transform data into a clean signal, extracting true characteristics and patterns.

• Utilized Support Vectors Regression (SVR) with different kernels and Neural Network as a predictive model.

• Designed and implemented an algorithm to optimize hyperparameter for each model, resulting in significant improvements in adjusted R-Squared and substantial reduction in Mean Squared Error (MSE) value.

• Conducted compressive analysis, comparing and visualizing model performance metrices, where Deep Neural Network with fourier transformed achived an out-of-sample prediction accuracy of 85% and MSE of 0.00234.

• Machine Learning For

Scientific Computing

• Linear Statistical Modeling

• Applied Stochastic Processes

• Econometrics and Machine

learning for Finance.

• Quantitative Risk

Management

• Linear Algebra

• Explainable Artificial

Intelligence (XAI)

• Large Language Model (LLM)

• Machine Learning(scikit -

learn, tensorflow)

• Stochastic Modeling

• Time Series Analysis

• Credit Risk Modeling

• Python(pandas, numpy,

matplot, colab)

• SQL

• Financial Derivatives

• Data Mining

• Optimizations

Research Assistant

Robert H. Smith School of Business

Aug 2022 – Dec 2023

• Manage dataset of 480 million multidimensional time series observations using Python and SQL in Jupyter Notebook.

• Analyze data distribution and relationships, including normality, outliers, and stationarity tests, as well as regression analysis.

• Prepare research reports, deliver presentations, and provide student support in data- mining, statistical inference, and course-specific topics. Tax Advisor Intern

Cash Campaign of Maryland

Feb 2019 – April 2019

• Analyzed financial data to identify potential deductions and credit for a portfolio of 20+ clients, resulting in a 10% reduction in their overall tax liabilities.

• Advised client on tax planning and compliance issues. SUDIP KHADKA

929-***-****

ad46f2@r.postjobfree.com

www.linkedin.com/in/sudip-khadka-25835812a

Baltimore, MD

SKILLS

Master of Mathematics, Finance

University of Maryland, College Park, MD

Aug 2022 – May 2024

Bachelor’s in Economics

Morgan State University, Baltimore, MD

Aug 2018 – May 2021

EDUCATION

Relevant Coursework

Contact this candidate