Hi, I'm Harshith Bandari

I'm a Data & Business Analytics student who turns raw, messy data into decisions people can act on. I like owning the full loop — cleaning data, validating it statistically, building predictive models, and translating the results into dollars and clear recommendations. Currently open to analytics internships.

• 🔭 Building end-to-end analytics projects across churn, NL-to-SQL, NLP, healthcare, and supply chain
• 📊 Past work: marketing analytics at IAMRATAN — funnel analysis, CAC/LTV, and attribution
• 🧠 Comfortable across the stack: SQL → Python → machine learning → Tableau dashboards
• 🎯 I care most about the "so what" — the business decision behind the number
• 🌱 Always sharpening my skills in stats, ML, and data storytelling

Languages & Databases

Data & Machine Learning

BI & Tools

SQL · Python · scikit-learn · Tableau

End-to-end pipeline that predicts churn and quantifies the revenue at risk.

• 💰 Identified ~$544K/year in retainable revenue from the high-risk segment
• 🎯 Churn model reaches ROC-AUC 0.85 (76% accuracy, 74% precision)
• 📊 Chi-square validation + risk-scoring + a live Tableau dashboard

Python · SQL · Streamlit · pytest

Turns plain-English business questions into safe, read-only SQL, then returns tables and charts.

• 🧩 Deterministic, schema-aware parser: 5 metrics × 8 dimensions, correct multi-table JOINs
• 🛡️ Read-only safety layer blocks destructive queries; auto-injects LIMIT
• ✅ 17 passing unit tests + an interactive Streamlit app

Python · scikit-learn · NLP

NLP pipeline that turns thousands of free-text reviews into decisions.

• 🎯 Sentiment model at 87.9% accuracy (TF-IDF + Logistic Regression, weak supervision from star ratings)
• 🧠 KMeans theme discovery + aspect-based sentiment to rank pain points
• 🚨 Auto-flags feature requests and urgent reviews into a triage queue + dashboard

Python · Classification

Healthcare classification model that predicts appointment no-shows and scores patients by risk, so clinics can target reminders where they matter most.

Python · Time-Series

Supply-chain project forecasting demand and optimizing inventory levels to balance stockouts against holding costs.