Indu Shahi

Data Scientist with 10+ years of experience designing and deploying machine learning and deep learning solutions, with hands-on work in recommender systems, demand forecasting, fraud/anomaly detection, and Generative AI / LLM applications (RAG, prompt engineering, BERT, transformers). Proficient in Python, PyTorch, TensorFlow, SQL, and Spark/Databricks across Azure, AWS, and GCP cloud environments. Skilled in designing A/B tests and measurement frameworks, building end-to-end ML pipelines, and partnering cross-functionally with engineering and product teams to deliver production-ready, data-driven solutions with a high degree of independence and ownership.

Data Scientist - SAFE Credit Union

(2022-12 - 2025-09)

• Designed and improved recommendation and propensity-scoring models combining customer behavior, transactional data, and contextual signals to support personalized product and marketing recommendations
• Applied Generative AI (RAG, prompt engineering) and BERT-based transformer models to generate sentiment insights and text-classification outputs from call-center feedback data, evaluating outputs against business-centric metrics
• Designed and ran A/B tests and uplift modeling for marketing campaigns; built measurement frameworks to translate experiment results into actionable model and business improvements
• Built and owned end-to-end ML pipelines combining data cleaning, feature engineering, model training and deployment in AWS SageMaker, partnering with engineering on production deployment and iteration
• Performed anomaly detection for credit card fraud transactions using Isolation Forest; developed churn prediction and scorecard risk models for finance using Python
• Carried out statistical analysis and ML modeling (ARIMA, Linear Regression) to forecast revenue, sales and new account openings; supported loan-approval analytics using house-price estimation models
• Curated and engineered datasets via data wrangling/ETL pipelines (Pandas, NumPy, SSIS) to support ML and AI applications; generated SQL reports/dashboards (Oracle, SSMS, SSAS, Power BI, Tableau)

Data Scientist - US Department of Veteran Affairs - Washington DC

(2022-05 - 2022-11)

• Built data and ML pipelines on Azure Databricks and Synapse, extracting and curating large datasets from Delta Lake using SQL and PySpark to support downstream ML and visualization applications
• Designed predictive and clustering models (Azure ML, Azure Machine Learning Studio) for income prediction, cancer-stage indication, and unstructured veteran-affairs data segmentation
• Conducted text analytics using transformers and BERT for sentiment analysis of call-center data; performed anomaly detection across VADIR, PATS-R and MPI datasets using Python, PySpark and SQL
• Performed EDA, statistical analysis and insight generation; translated findings into Power BI and Tableau dashboards for cross-functional, non-technical stakeholders across departments and regions
• Collaborated with engineering and program teams in an Azure cloud, fast-paced environment, communicating technical findings clearly while operating with high independence

Data Scientist - AT&T - Dallas TX

(2021-11 - 2022-04)

• Applied singular value decomposition (SVD)-based collaborative filtering in Python and Scala to recommend cost-optimized Azure virtual machine configurations — core recommender-system experience using user/resource utilization signals
• Built data engineering pipelines extracting resource, application and utilization data from Azure Delta Lake; applied PySpark UDFs to process CPU, IOPS and disk-utilization data at scale
• Implemented scalable demand-forecasting solutions on cloud platforms (AWS, GCP) and big-data technologies (Spark, Hadoop) handling millions of data points daily; evaluated models using RMSE and XGBoost
• Built and maintained dashboards using Databricks notebooks to communicate model outputs to engineering and business stakeholders

Data Scientist - SMUD - Sacramento CA

(2018-06 - 2021-10)

• Conducted customer segmentation and propensity-score matching on unstructured data using Python and SageMaker; classified client profiles using KNN to support personalization use cases
• Built and validated regression, Random Forest and SVM models for utility consumption forecasting; performed sentiment analysis on call-center data using NLP (NLTK, SpaCy, Gensim)
• Delivered statistical analysis, data wrangling and EDA using Pandas/NumPy/Python and R; built visualizations and anomaly-detection dashboards using Tableau, Power BI and GGPlot2

Data Scientist - ARCO - Sacramento, CA

(2017-01 - 2018-06)

• Led development of large-scale demand-forecasting models using statistical and machine-learning techniques (ARMA/ARIMA), improving forecast accuracy and operational efficiency
• Applied uplift modeling to identify customers likely to respond to promotions/discounts; built logistic-regression fraud-detection model in Python
• Performed data modeling, wrangling and EDA on sales and inventory data in AWS; built sales/profit dashboards and visualizations using Tableau and Power BI

Data Analyst - Friends of Nepal - New Jersey, USA

(2016-09 - 2017-01)

• Performed data wrangling, EDA and logistic-regression modeling in Python to analyze insurance-claim data and predict maternal-mortality reduction outcomes
• Built a KNN model in R for population segmentation; predicted insurance sales using STATA with visualization in Tableau and Power BI
• Automated reporting in Excel; authored a research report on the contribution of insurance to reducing maternal mortality

Data Analyst/Consultant - Katahira International Japan - Kathmandu, Nepal

(2014-11 - 2015-07)

• Extracted, cleaned, modeled and visualized casualty-insurance data using SQL (Oracle), Python and Tableau
• Designed a machine-learning logit model in STATA to classify fraudulent insurance claims; estimated claim size/frequency using AR, MA and ARMA time-series models
• Presented findings and recommended best-performing classification models to stakeholders based on tested accuracy and validation

Data Analyst/Consultant - Social Security Fund Secretariat - Kathmandu, Nepal

(2012-11 - 2014-10)

• Reviewed comparative social-insurance and mortality models across Thailand, South Korea, India and Tunisia
• Built ML time-series models (AR, MA, ARIMA, OLR) in R/STATA to forecast revenue, premium collection and claim size; performed feature engineering and advanced analytics
• Authored and secured government approval for an independent policy report on social insurance for the Government of Nepal

Actuarial/Data Analyst - National Life Insurance Co., Ltd. - Kathmandu, Nepal

(2009-02 - 2012-10)

• Taught advanced statistics, econometrics and machine-learning concepts (regression, classification, optimization) to undergraduate students
• Provided advanced analytical support, sales forecasting and training materials