Data Scientist · Fresher · 0 yrs

Ananya Deshpande

Data Scientist (Entry-Level)

Professional summary

Final-year Statistics & Computer Science graduate who turns messy data into clear decisions using Python, R and SQL. Comfortable across the workflow — exploratory analysis in pandas, hypothesis testing, machine-learning models in scikit-learn, and crisp data visualization in Tableau. Analytically curious and rigorous about statistics, demonstrated across a forecasting capstone, a churn-prediction project and a Kaggle medal.

0.88

Churn model ROC-AUC

11%

Capstone forecast MAPE

Top 8%

Kaggle competition rank

Skills

Languages

PythonRSQLBash

ML & stats

scikit-learnstatisticsmachine learninghypothesis testingA/B testingregressiontime-series

Data wrangling

pandasNumPydplyrJupyterfeature engineering

Visualization & BI

Tableaumatplotlibseabornggplot2Power BI

Work experience

Data Science Intern · Lumio Analytics

May 2025Aug 2025

Remote

  • Built a churn-prediction model in scikit-learn (gradient boosting) reaching 0.88 ROC-AUC, flagging at-risk accounts 3 weeks earlier than the existing rules engine.
  • Automated a weekly KPI pipeline in Python + SQL that replaced a manual spreadsheet, saving the analytics team ~6 hours/week.
  • Ran an A/B test on two onboarding emails (n = 18k), establishing statistical significance (p = 0.013) and a 9% lift in activation.
  • Designed 4 Tableau dashboards for the growth team, cutting recurring ad-hoc data requests roughly 40%.

Featured projects

MAPE 19% → 11%
DemandCast — retail demand forecasting (capstone)

Time-series forecasting system for a mid-size grocery chain, built as the degree capstone.

  • Engineered features from 3 years of sales data in pandas and benchmarked SARIMA against gradient-boosted trees, cutting forecast MAPE from 19% to 11%.
  • Shipped an interactive Tableau dashboard so category managers could explore forecasts by store and SKU.
Pythontime-seriesscikit-learnTableau
0.91 fraud recall
FraudLens — transaction anomaly detection

Imbalanced-classification study on a public payments dataset.

  • Handled severe class imbalance with SMOTE and cost-sensitive learning, lifting recall on fraud cases to 0.91 at fixed precision.
  • Documented the statistical evaluation (precision-recall, calibration) in a reproducible R Markdown report.
PythonRmachine learningstatistics

Education

B.Sc. (Hons.) Statistics & Computer Science

2026

Fergusson College (Savitribai Phule Pune University), Pune

CGPA 9.1/10 · Coursework: Probability, Inference, Regression, Machine Learning, Databases, Data Visualization

Certifications

IBM Data Science (Professional Certificate)

Coursera · 2025

Tableau Desktop Specialist

Tableau · 2025

See the work in 3D

Explore Ananya's interactive WebGL portfolio — projects, skills and a way to get in touch.