Looking for a job

Shalini Rajanna

@shalinirajanna

Senior Evidence Data Scientist at Microsoft who cut product analysis cycles from 18 days to 9 with reproducible Python and SQL workflows.

United States

Message

At Microsoft, I designed retrospective product-impact studies for cloud and enterprise teams, defining hypotheses, endpoints, cohorts, and analysis plans for longitudinal usage data. I built reproducible Python and SQL workflows that cut analysis cycle time from 18 days to 9.

I evaluated data quality, provenance, missingness, selection bias, and confounding before analysis, helping research partners distinguish measurement gaps from meaningful product effects. My work also applied causal inference and machine learning to explain adoption drivers and forecast customer behavior.

At Rockstar Games, I developed player-behavior studies for live game features and built pipelines for longitudinal player cohorts. Predictive modeling raised recall from 62% to 78% on a held-out player segment.

At Bank of America, I analyzed customer and transaction behavior for risk and servicing research, building models that improved fraud-case capture from 79% to 86%. Earlier, as a Graduate Research Assistant in the DAIR lab at UNC Charlotte, I implemented Python and TensorFlow workflows for machine learning experiments.

Experience

Work history, roles, and key accomplishments

Current

Senior Applied Data Scientist

Current

Microsoft

Mar 2021 - Present (5 years 7 months)

Designed retrospective product-impact studies for cloud and enterprise product teams, defining hypotheses, endpoints, cohort rules, and analytical plans. Built reproducible Python and SQL workflows, reducing analysis cycle time from 18 days to 9 days.

Python SQL regression Hypothesis Testing Uncertainty estimation Causal Inference Machine Learning Azure Databricks Apache Spark Azure Machine Learning

Senior Machine Learning Engineer

Rockstar Games

Sep 2020 - Mar 2021 (6 months)

Developed player-behavior studies for live game features, defining measurable outcomes and comparison groups from high-volume event and session data. Built Python and SQL pipelines for longitudinal player cohorts, raising recall from 62% to 78% on a held-out player segment.

Python SQL predictive modeling Classification Feature Engineering Data Pipelines

Data Scientist

Bank of America

Feb 2014 - Sep 2020 (6 years 7 months)

Analyzed customer and transaction behavior for risk and servicing research, defining cohorts, derived variables, and outcome measures. Built predictive models and analytical dashboards that improved fraud-case capture from 79% to 86% while preserving analyst review capacity.

SQL Python regression Classification Clustering Time Series PySpark Pandas scikit learn

Graduate Research Assistant

UNC Charlotte

Sep 2012 - Dec 2013 (1 year 3 months)

Formulated research questions and prepared experimental datasets for machine learning studies in the DAIR lab. Implemented feature extraction, classification, and model evaluation workflows in Python and TensorFlow, increasing held-out accuracy from 68% to 76%.

Python TensorFlow Feature Extraction Classification Model Evaluation