I'm a data analyst and ML engineer currently pursuing my Master's in Business Analytics at WPI. My work sits at the intersection of statistical rigor and applied engineering — I build things that analyze, predict, and explain. When I'm not running regressions or training models, I'm building the tools and products that make data accessible to people who need it.

I care about work that means something. My capstone identified three systemic failures in US hospital penalty policy. My side projects range from voice cloning backends to no-code AutoML platforms. The common thread is the same: find a real problem, understand it deeply, build something that helps.

WPI BUS596 capstone. Cross-sectional OLS regression across 11 merged CMS public-use files and 2,833 U.S. acute care hospitals. Identified three distinct failure modes in federal hospital penalty programs — all findings significant at p < 0.001.

Includes a self-contained interactive research site with a live hospital explorer across all 2,833 hospitals, tabbed charts, and a fullscreen conference poster mode.

No-code automated data analysis platform. Upload a CSV or Excel file — get instant EDA, cleaning diagnostics, ML feasibility scoring, and structured insights without writing a single line of code.

• 6-service FastAPI backend: Parser, Cleaning, EDA, Scope, Insight, AutoML
• AutoML pipeline running 5 simultaneous models with ranked comparison and confidence scoring
• Interactive cleaning panel with apply, preview, and undo operations
• Supabase Postgres for persistent public report sharing and stateless OTP auth

Self-hosted voice synthesis backend using Coqui XTTS v2. Clones and consistently reproduces an assistant-style voice from a single short WAV reference clip — no training required. Automatic GPU/CPU detection with cached conditioning latents for low-latency inference. Production-ready REST API with /speak, /health, and /info endpoints.

End-to-end ML pipeline for income classification on the Adult UCI dataset. Feature engineering, hyperparameter tuning, k-fold cross-validation, and model deployment as a managed online endpoint on Azure ML Studio. Optimised AUC-ROC using ensemble boosted decision trees.

Kaggle survival prediction. Feature engineering across title extraction, family size, and cabin deck variables. Ensemble of Random Forest and XGBoost with k-fold cross-validation. Final score: 0.78708 accuracy · Top 35% globally.

Languages   Python · SQL · JavaScript · HTML/CSS

ML & Data   Pandas · NumPy · SciPy · scikit-learn · statsmodels · XGBoost · PyTorch · Coqui XTTS

Engineering   FastAPI · Next.js · React · Docker · Supabase · Vercel · Render · Azure ML Studio

Analytics   Power BI · DAX · Google Analytics 4 · Exploratory Data Analysis · OLS Regression

Worcester Polytechnic Institute — MS Business Analytics  ·  GPA 4.0 / 4.0  ·  Jan 2025 – Dec 2026

Relevant Certifications

• Machine Learning Specialization — DeepLearning.AI & Stanford University (Mar. 2026)
• Machine Learning Pipelines with Azure ML Studio — Coursera (Mar. 2026)
• Google Analytics 4 Certification (Apr. 2025)
• Google Ads Search Certification (Apr. 2025)