Architected and optimized a centralized Enterprise Finance Data Lake, ingesting transactional and ledger data from diverse ERPs (SAP, Oracle)
using Python, SQL, Spark, Hadoop, and Kafka, resulting in a 30% faster data processing pipeline for predictive analytics.
Formed and implemented feature engineering strategies for predictive modeling, NLP-based insights, and programmatic data augmentation,
improving entity extraction, sentiment analysis, and model accuracy by 18%.
Developed metadata-driven ETL workflows across cloud and on-premise sources, harmonizing heterogeneous schemas and reducing
missing/null errors by 40% for downstream analytics.
Orchestrated interactive Tableau dashboards integrating Snowflake and batch/streaming data, enabling real-time KPIs, trend analysis, anomaly
detection, and 25% faster business decision-making for finance stakeholders.
Deployed customized retrieval pipelines and rule-based scenario simulations to integrate structured and unstructured financial data, enabling
context-aware reporting and stress-testing across accounting, compliance, and operations.
Formulated and deployed predictive models for financial forecasting using Python and ML frameworks, integrating with AWS (S3, SageMaker,
Lambda) and advanced time series methods (TFT, Prophet), achieving accurate 6-month revenue and demand predictions with MAE < 5%.
Containerized ML workflows with Docker on Kubernetes clusters, ensuring scalable, reproducible model training and inference pipelines,
reducing infrastructure overhead by 40%.
Applied AI ethics, model explainability (SHAP, LIME), and manual drift monitoring, identifying 1,500+ biased predictions and maintaining
compliance with regulatory standards.