- Designed and enhanced backend ETL pipelines by leveraging gRPC for service communication and .Net Framework and Java servers to integrate eFront (alternative investment data) into Aladdin, enabling institutional clients to manage unified portfolios and positively impacting $1.5T in managed assets.- Built a Python-based reconciliation server to validate and synchronize financial data between eFront and Aladdin, reducing synchronization errors by 70% and achieving 99.8% ETL accuracy. Implemented automated data validation checks and hash comparisons to ensure data consistency and reliability for critical investment decisions.- Spearheaded enhancements to Aladdin’s whole-portfolio solution by optimizing data flow, extending APIs, and improving data integration logic. These improvements drove a 15% increase in client retention, accelerated product adoption, and contributed to $250M+ in annual revenue.

• Supported Payment Risk & Data Science under the Marketplace Payment team- Conducted advanced fraud detection analyses using SQL, Python (Pandas, Scikit-learn), and Power BI, resulting in a 25% reduction in fraudulent transactions and improving the integrity of marketplace payments.- Built and optimized predictive models, including random forest and ridge regression, to improve fraud detection accuracy by 35% across 1.2M vendors, safeguarding Walmart’s marketplace ecosystem.- Designed and executed exploratory data analyses to identify risk patterns, reducing false positives by 20% and saving $5M annually in disputed transactions.- Enhanced data pipelines to support real-time analytics, improving time-to-insight by 30% and enabling actionable fraud prevention strategies.

• Analyzed call center data from the data warehouse by using a set of SQL queries and implementing exploratory and predictive analytics (including causal inference and time series modeling) using tools, including Python (Scikit-learn, Pandas, NumPy), PowerBI, and SPSS. • Predicted customer sentiment scores of each agent by tenure and location and identified features that have the biggest impact on average call handled time using machine learning such as random forest (bagging), linear regression (ridge), and clustering.

• Worked with Inventory Replenishment and Planning (IPR) team taking on duty for data wrangling and visualization using Google Big Query, Python, and Tableau to build a set of dashboards with KPIs that improve inventory planning and replenishment process with cross-functional efforts• Involved with end-to-end development of multiple supply-chain management projects, including new stores inventory tracker, Ice Melt seasonal tracking, PO tracker, to help stakeholders make strategic fulfillment decisions within the organization to minimize the out-of-stock instances of products and to maximize just-in-time sales

• Worked with Space Optimization team (SPOT) and utilized GCP to explore, extract, transform, and load (ETL) diverse datasets from both online and in-store databases using a set of SQL queries. Using Python, built recommender system upon transactional and customer behavioral data to identify substitutable product groups with unsupervised machine learning techniques • Dealt with sample size issues and imbalanced labels with imputation and oversampling techniques and set up scalable end-to-end machine learning pipeline (supervised learning – linear/ridge/Poisson regression) to predict demand transference among substitutable products and provided the client with actionable insights to improve in-store assortments to minimize $1.4B lost sales due to shelf-out

• Analyzed the company's financial status and created financial models• Created an investor deck by conducting comprehensive and in-depth research on the select industries

Primary author: https://www.ncbi.nlm.nih.gov/pmc/articles/PMC6656589/ https://diabetesjournals.org/diabetes/article/67/Supplement_1/2031-P/55101/Low-Dose-FGF-21-Administration-Decreases-Obesity