About

Improved demand forecasting accuracy by 15% through trend and scenario analysis, and outperformed volume prediction baselines by 14.5–20.6% with Bayesian–ARIMA optimization across 275 time series with full uncertainty quantification. Evaluated 8+ GenAI and Agentic AI architectures for production workflow automation. MS Data Science from Seattle University with applied experience in forecasting, model optimization, explainable analytics, and building automated data pipelines in Python, R, and SQL.

Experience2

1. R

   Data Science Intern

   Jan 2024 — Jun 2024

   RentPrompts

   Evaluated 8+ GenAI and Agentic AI architectures for automated workflow use cases by benchmarking accuracy, latency, and cost trade-offs on Hugging Face and GPU-enabled environments. Launched 3 optimized models into the production platform by designing offline evaluation experiments measuring output quality and processing efficiency. Shortened experimentation cycles by prototyping a reusable data ingestion pipeline that standardized raw inputs into model-ready formats.

2. D

   Data Analyst Intern

   Jan 2023 — Jun 2023

   DisplayFort

   Increased operational conversion rate by 18% by designing a recommendation model with customer segmentation in Python and Scikit-Learn. Cut ad hoc data preparation time by 40% by building automated SQL pipelines. Improved quarterly demand projection accuracy by 15% through scenario analysis and simulation on 12 months of operational data.

Projects2

• Forecasting Optimization & Explainable Analytics (Capstone)

  Outperformed naive volume forecasting baselines by 14.5% and 20.6% using an Empirical Bayesian–ARIMA ensemble with parameter optimization. Engineered a scalable data pipeline merging multi-source operational data into a unified analytical layer, applying feature engineering and dimensionality reduction.

  • Python
  • Scikit-learn
  • PyMC
  • Streamlit

• Predictive Classification & Model Selection (Federal Survey, 30K+ Records)

  Achieved 88% accuracy across 30,000+ records by implementing and comparing multiple classification algorithms. Engineered 60+ features from noisy behavioral data and reduced prediction error by 21% through iterative model selection.

  • R
  • Random Forest
  • XGBoost

Education2

• SU

  Master of Science in Data Science

  2024 — 2026

  Seattle University

• RG

  Bachelor of Technology in Computer Science

  2019 — 2023

  Rajiv Gandhi Proudyogiki Vishwavidyalaya