Hi, I'm Abhi
Data Scientist
I specialize in applied machine learning and statistical modeling, focused on building interpretable models and translating analytical findings into compelling visual narratives.
I specialize in applied machine learning and statistical modeling, focused on building interpretable models and translating analytical findings into compelling visual narratives.
I'm a data scientist with a background in statistics and a strong interest in machine learning, forecasting, and optimization. I enjoy building models that help explain complex systems, whether that's predicting future outcomes, identifying patterns in noisy data, or uncovering the factors that drive real world decisions. Most of my projects begin with a question rather than a model. I enjoy exploring datasets, testing assumptions, engineering meaningful features, and translating quantitative findings into clear visual stories. Outside of work, I use this space to explore problems that genuinely interest me and to document the methods, successes, and lessons learned along the way.
B.Sc. in Data Science
The University of Texas at Dallas, 2022
Machine Learning Analyst
Raas Infotek, 2024–Present
Binary Classification | Merging Multi-Source Data → Feature Engineering → Predictive Modeling
Building a classification model to predict loan default risk using a merged dataset of 50K+ customer, loan, and credit bureau records. Engineered domain-specific features including loan-to-income ratio, delinquency ratio, and average days-past-due per delinquency. Applied VIF analysis and Information Value scoring to reduce 33 features to a focused set of 10 predictors. Logistic regression baseline achieved 96% accuracy with 85% precision on default cases.
End-to-end regression modeling | EDA → Feature Engineering → Model Evaluation
Built a regression model to predict California housing prices using demographic and geographic data from ~20K+ records. Conducted exploratory data analysis and statistical validation (OLS) to identify significant predictors. Improved model performance by incorporating spatial features, increasing R² from ~0.50 to ~0.60 and reducing RMSE by ~8%. Identified median income and location as key drivers of housing prices.
ML Web App | Clustering → NLP Search → Streamlit Deployment
Built a personalized "For You" page for datathon competitors in under 24 hours, selected from 1,000+ applicants. Implemented team recommendation using clustering on competitor attributes (experience, university, age bracket). Developed an NLP search engine to surface relevant workshops from user-inputted keywords via a logistic regression classifier. Deployed as a fully interactive Streamlit app with data visualizations of competitor demographics and LinkedIn internship data.
Exploratory Data Analysis | Data Cleaning → Feature Exploration → Insight Generation
Conducted EDA on ~20K Airbnb listings to analyze pricing dynamics and neighborhood trends.
Identified location and room type as primary drivers of price, with central districts commanding premium rates.
Explored host-level attributes and review behavior to uncover patterns in customer engagement.
Python, SQL, R, MATLAB
Regression, Classification, Ensemble Methods, GLMs, A/B Testing, Hypothesis Testing, Model Evaluation (RMSE, R², F1)
EDA, Feature Engineering, Data Cleaning, ETL Pipelines, Data Validation, SQL Optimization
Tableau, Power BI, Plotly, Matplotlib, Seaborn
Git, Jupyter, Google Colab, VS Code, Streamlit
SQL Server, Relational Databases, Data Modeling