Deepa

Senior Data Scientist with 7+ years of experience in machine learning, marketing analytics, and revenue optimization across retail, e-commerce, and payments, working extensively with large-scale datasets ranging from 500K to 1M+ records across sales, inventory, and customer domains. Demonstrated expertise in designing, developing, and deploying time series forecasting solutions (ARIMA, Prophet), including trend and seasonality analysis, rolling-window forecasting, and historical back-testing to support revenue management and demand planning. Strong proficiency in Python (Pandas, NumPy Scikit-Learn, SciPy, Statsmodels, XGBoost, LightGBM) to build production-grade predictive models and perform advanced feature engineering on transactional and behavioral datasets.

Advanced proficiency in SQL with deep hands-on experience in Amazon Redshift and Snowflake, including complex joins, CTEs, window functions, performance tuning, and warehouse optimization for analytics workloads. Expert in customer segmentation (K-Means, RFM Analysis), pricing and revenue optimization, NLP and LLM prompt engineering, and A/B testing to drive data-informed marketing and merchandising strategies. Hands-on experience operationalizing analytics using Databricks, Amazon Redshift, AWS S3, and Google Colab, following CI/CD best practices with Airflow and Control-M.

Extensive experience building enterprise data warehouse solutions, including ETL pipeline design, multi-source data consolidation, and centralized reporting architectures in Amazon Redshift and Snowflake. Adept at data storytelling and visualization using Power BI, Tableau, and Amazon QuickSight. Strong communicator with proven ability to mentor junior team members and partner cross-functionally with marketing, finance, and technology stakeholders.

Senior Data Scientist - KeHE Distributors

(2023-07)

• Led development of pricing and revenue optimization models using Python (Pandas, Scikit-Learn) in Google Colab to analyze historical sales and margin data across 1M+ SKU-level records, enabling data-driven pricing strategies that improved decision-making efficiency by 18%.
• Built and deployed time series forecasting models (ARIMA, Prophet) on multi-year retailer sales and inventory data, reducing forecast error by 15% and empowering revenue management to proactively adjust demand planning and promotional calendars.
• Architected the Retailer Analytics dataset from legacy NTS systems by designing structured ETL pipelines and consolidating multi-source data into a centralized Amazon Redshift warehouse, establishing a single source of truth for enterprise reporting across sales, inventory, and pricing domains.
• Engineered and optimized scalable SQL queries across 1M+ records in Amazon Redshift, leveraging CTEs, window functions, and partitioning strategies to streamline data extraction and validation, reducing ad-hoc reporting turnaround time by 30%.
• Designed interactive Power BI dashboards with advanced DAX measures tracking sales performance, inventory health, and pricing KPIs; automated reporting workflows using DAX and Python, improving load times by 20% and eliminating ~9 hours of manual effort per week.
• Spearheaded a marketing NLP initiative by applying LLM prompt engineering (GPT-4 API) to analyze customer feedback and product review data, extracting sentiment trends and category-level insights that informed targeted marketing campaigns and improved promotional content strategy.
• Developed a customer segmentation model using K-Means clustering and RFM (Recency, Frequency, Monetary) analysis in Python (Scikit-Learn) to classify 500K+ customer records into high-value behavioral cohorts, enabling targeted marketing strategies that improved campaign conversion rates by 25%.
• Built a demand forecasting pipeline using XGBoost and LightGBM by engineering 30+ temporal and behavioral features from transactional data, achieving a 12% improvement in demand prediction accuracy over baseline models.
• Implemented feature engineering pipelines incorporating lag variables, rolling averages, seasonality flags, and calendar-based indicators using Pandas and NumPy to enhance model inputs for forecasting and pricing optimization workflows.
• Conducted A/B testing frameworks to evaluate promotional pricing strategies across product categories, translating statistical insights into actionable recommendations for marketing and finance teams.
• Performed comprehensive back-testing and rolling-window validation of forecasting outputs, evaluating model accuracy and stability using metrics such as RMSE, MAPE, and MAE to ensure business readiness prior to production deployment.
• Built scalable ETL and data ingestion pipelines using Airflow and Control-M to orchestrate daily data refreshes from multiple source systems into Amazon Redshift, ensuring data timeliness and reliability for downstream analytics.
• Partnered closely with revenue management, marketing, and finance stakeholders to define analytical approaches and convert model outputs into actionable operational insights; authored technical documentation in Confluence and JIRA to support maintainability and knowledge transfer.
• Mentored junior data analysts on Python-based modeling workflows, SQL optimization techniques, and Power BI dashboard development, accelerating onboarding timelines and improving team output quality.
• Managed version control and collaborative development workflows using Git and followed Agile/Scrum practices to ensure consistent delivery cadence and cross-functional stakeholder alignment.

SDE - Data Scientist 1 - Amazon - India

(2019-09 - 2021-12)

• Developed and deployed ML models (Random Forest, XGBoost, Logistic Regression) in Python (Scikit-Learn, Pandas) by engineering features from 1M+ transaction records to flag high-risk payments, reducing fraud by 20% and improving annual savings by 5%.
• Led a cross-functional marketing analytics initiative by building customer conversion and purchase propensity models using Python and SQL, analyzing buyer behavior across 6 global marketplaces to identify high-value segments, driving a 12% lift in campaign conversion rates.
• Designed and maintained Amazon Redshift-based ETL pipelines powering real-time BI dashboards by orchestrating data ingestion from multiple sources, contributing to a 16% reduction in fraud rates and an 8% decrease in Account Take Over incidents.
• Built A/B testing frameworks and Tableau dashboards to measure marketing campaign effectiveness and promotional pricing strategies, translating analytics into executive-level insights that informed budget allocation.
• Performed extensive feature engineering on transactional datasets, creating lag variables, rolling aggregations, time-based behavioral indicators, and interaction features using Pandas and NumPy to enhance fraud detection model performance.
• Applied customer segmentation techniques (K-Means, RFM) using Scikit-Learn to classify buyer cohorts across global marketplaces, enabling personalized marketing strategies and targeted promotional campaigns.
• Implemented predictive analytics pipelines for payment risk scoring using XGBoost and LightGBM, incorporating model explainability techniques (SHAP values) to ensure transparency and regulatory compliance.
• Engineered scalable SQL queries and stored procedures in Amazon Redshift to support ad-hoc analytics, fraud trend monitoring, and automated reporting across payments and marketplace domains.
• Developed real-time monitoring dashboards using Amazon QuickSight and Tableau to track fraud KPIs, payment success rates, and customer behavior anomalies, supporting rapid incident response and proactive risk mitigation.
• Conducted statistical hypothesis testing and A/B experimentation using SciPy and Statsmodels to evaluate the effectiveness of fraud prevention policies and marketing interventions across multiple treatment groups.
• Automated data extraction, transformation, and reporting workflows using Python scripts and Airflow orchestration, reducing manual effort by 25% and improving data freshness for downstream analytics.
• Collaborated with cross-functional teams across payments, marketing, and product to define KPIs, align on analytical priorities, and present data-driven recommendations to senior leadership.
• Mentored junior analysts on Python-based modeling workflows, model evaluation techniques (precision, recall, AUC-ROC), and SQL best practices, accelerating onboarding timelines and improving team output quality.
• Managed code repositories using Git and tracked project deliverables through JIRA and Confluence, following Agile sprint-based delivery practices.

Student Research Assistant - University of Connecticut

(2022-02 - 2023-04)

• Built Power BI (2022–2023) dashboards tracking revenue, expenses, and budget variances across departmental accounts, reducing financial report preparation time by 20% and improving transparency for administrative leadership.
• Monitored financial KPIs and performed variance analysis using DAX measures and SQL queries to identify cost optimization opportunities, improving budget visibility and enabling proactive fiscal planning.
• Developed end-to-end predictive models across annuity contracts, fuel prices, and salary benchmarking datasets using Python (Scikit-Learn, Pandas) and SAS, driving insights that contributed to a 20% churn reduction in annuity retention modeling.
• Applied machine learning techniques (Decision Trees, Random Forest, Regression) and time series forecasting to predict customer retention trends, fuel price trajectories, and compensation benchmarks across diverse analytical use cases.
• Performed comprehensive feature engineering, model evaluation, and hyperparameter tuning using GridSearchCV and cross-validation strategies to maximize prediction accuracy and generalization across multiple forecasting scenarios.
• Conducted business impact analysis on predictive model outputs, translating statistical results into clear, actionable recommendations for faculty and administrative stakeholders.
• Developed a collaborative filtering Movie Recommendation System using K-Nearest Neighbors (KNN) in Python (Scikit-Learn, NumPy), performing exploratory data analysis and similarity modeling to generate personalized recommendations.
• Designed a prototype UI to simulate user interaction and recommendation delivery, demonstrating end-to-end ML pipeline capabilities from data preprocessing through model serving.
• Automated recurring administrative data extraction and reporting tasks using Python scripts and Excel (Pivot Tables, VLOOKUPs), reducing manual effort and improving data accuracy for departmental operations.
• Presented analytical findings and dashboard insights to cross-functional academic and administrative teams, supporting data-driven decision-making for budget allocation and resource planning.

Data Analyst Intern - BHEL (Bharat Heavy Electricals Limited) - India

(2019-01 - 2019-08)

• Tracked and analyzed technical and financial data using Excel (Pivot Tables, VLOOKUPs, Conditional Formatting) to generate comprehensive reports on machinery performance, procurement costs, and service utilization trends.
• Negotiated pricing for machinery services by leveraging data-driven cost analysis, ensuring service continuity and enhancement while slashing procurement costs by 6%.
• Built automated reporting templates in Excel to consolidate multi-source operational data, reducing manual report generation time and improving consistency of deliverables for engineering and finance teams.
• Performed exploratory data analysis on equipment maintenance and performance logs to identify patterns, anomalies, and cost optimization opportunities across manufacturing operations.
• Collaborated with engineering and procurement teams to present analytical findings and support data-driven negotiations, contributing to improved vendor management and cost efficiency.

M.S. - Business Analytics and Project Management - University of Connecticut (2023-05)

B.E. - Electronics and Communications Engineering - Osmania University (2019-05)