Data-Driven Decision Making

Atul Sharma

Data Scientist / Data Analyst focused on building interpretable machine learning models for risk-sensitive and decision-critical problems.

My work emphasizes data quality, model evaluation, and explainability—ensuring that analytical outputs can be trusted, validated, and used in real-world decision-making.

📊 Data
🤖 Model
📈 Insight
✅ Decision

Featured Projects

Interpretable Credit Risk Modeling

LendingClub loan default prediction using Logistic Regression, Decision Trees, and XGBoost. ROC-AUC ~0.70 with feature importance and explainability.

JPMorgan Quantitative Research (Forage)

Estimated Probability of Default (PD), FICO quantization, and monotonic risk bucketing using Python and dynamic programming.

British Airways – Data Science Simulation

Web-scraped customer reviews, performed EDA, and modeled buying behavior.

Deloitte & Bank of America Simulations

Business analytics, Tableau dashboards, SQL analysis, and client-facing financial insights.

Other Applied Data Science Projects

LLM Quantitative Evaluation

Designed quantitative metrics to evaluate LLM outputs, focusing on consistency, accuracy, and performance trade-offs.

British Airways Customer Analytics

Customer sentiment analysis and behavioral insights using scraped reviews and exploratory data analysis.

Social Media Data Analysis

Exploratory and statistical analysis of social media engagement data to identify trends and user behavior.

Natural Gas Price Analysis

Time-series analysis and visualization of natural gas pricing data with focus on trend detection and volatility.

Skills

Python, SQL, Machine Learning, Credit Risk (PD), Statistics, Data Visualization (Tableau, Power BI), ETL, Data Quality, Decision Science

Education

M.S. Big Data Science — University of the Potomac (GPA 4.0)
B.A. LL.B. — Maharishi Markandeshwar University

F-1 Student | STEM OPT Eligible