Data Analyst · ML Engineer · Business Intelligence
Bengaluru, Karnataka, India

I build data systems that drive decisions — predictive ML models, analytics pipelines, and BI dashboards on large-scale, real-world datasets.

What sets me apart: I've analyzed real production data on factory floors at Elecon Engineering and Jinal Engineering — surfaced process inefficiencies that caused 12–15% excess cycle time and presented findings to engineering teams. That operational grounding gives me the domain context to separate insight from noise.

Recent highlights:

• 🏏 XGBoost classifier on 278K+ IPL records → 77% accuracy, AUC-ROC 0.83 — deployed as a live Streamlit win-probability dashboard with full ETL → feature engineering → tuning → deployment pipeline
• 🫀 CDC mortality risk predictor on 433K+ Americans → AUC-ROC 0.8256, 85% Recall — Streamlit app with Groq-powered AI clinical assistant (LLaMA 3.3 70B)
• 📊 End-to-end SaaS analytics pipeline → identified 3 churn drivers, computed MRR, CAC, LTV & churn rate via SQL cohort analysis, delivered multi-view Tableau dashboard

🎓 Currently: Data Science & Generative AI — Great Lakes Institute of Management (ML · Deep Learning · NLP · LLMs)

💼 Open to: Data Analyst · ML Engineer · BI Analyst · Data Scientist — Bengaluru or Remote

Languages

ML & AI

Visualization & BI

Databases & Tools

Python · XGBoost · Scikit-Learn · Streamlit · Pandas

• Trained an XGBoost model on 278,000+ ball-by-ball records across 10 IPL seasons
• Engineered 6 predictive features: dynamic run rate, wickets-in-hand, required run rate, pressure index, match phase, and venue factor
• Achieved 77% accuracy and AUC-ROC of 0.83, beating logistic regression baseline by 11 points
• Deployed a real-time Streamlit dashboard for live win-probability simulation

Python · SQL · Tableau · Pandas · Seaborn

• Built an enterprise-scale SaaS dataset (5,000+ records) modeled on real CRM, billing & usage-log structures
• Surfaced 3 primary churn drivers through EDA — revealing revenue-risk signals previously undetected
• Computed MRR, Churn Rate, CAC & LTV using SQL cohort analysis
• Designed a multi-view Tableau dashboard for Sales, Product & Customer Success teams

Python · XGBoost · Streamlit · Groq / LLaMA 3.3 70B · Plotly · SHAP · Scikit-Learn

• Built a binary classification system trained on 433,000+ Americans from the CDC BRFSS 2023 survey to identify individuals at HIGH mortality risk using only self-reported behavioral & demographic data — no lab tests required
• Engineered 4 domain-driven features (comorbidity_count, age_risk_tier, health_burden, ses_score) following strict zero data-leakage discipline — caught and fixed a 98.88% phantom-accuracy bug caused by target column leakage
• Achieved AUC-ROC: 0.8256 and 85% Recall at threshold 0.40 with a tuned XGBoost model (RandomizedSearchCV, 50 iterations × 3 folds); outperformed Logistic Regression, Random Forest, LightGBM & Neural Network baselines
• Deployed a full-stack Streamlit app featuring local auth (SHA-256), risk gauge dial, SHAP explainability charts, radar/bar factor charts, and a Groq-powered AI clinical assistant (LLaMA 3.3 70B) for plain-language prediction explanations
• Applied CRISP-DM across 12 documented steps with 7-strategy logical imputation on 433K rows and SHAP TreeExplainer for model transparency

• 🏫 B.E. Mechanical Engineering — GCET, Anand, Gujarat (2021–2024)
• 🏫 Diploma in Mechanical Engineering — N.G. Patel Polytechnic (2018–2021)
• 📜 Data Science & Generative AI Program — Great Lakes Institute of Management (Oct 2025 – Present)
  • Machine Learning · Deep Learning · NLP · Generative AI & LLMs · Statistical Modeling

Open to Data Analyst, ML E