Implemented Medallion Architecture in Databricks: cleaned raw data (bronze), transformed and aggregated (silver), and delivered business-ready datasets (gold).
Conducted Exploratory Data Analysis (EDA) using PySpark to identify correlations between settlement price, wind, rainfall, and lake storage, boosting forecasting accuracy by 15%.
Trained and optimized a Temporal Fusion Transformer (TFT) model using past data, covariates, and future data, with hyperparameter tuning to optimize performance.
Deployed the best-performing model using GoLive capabilities within the Power and Utility Framework, enhancing forecast reliability by 20%.
Built a REST API displaying 3 key features: rain, wind, and price forecasts to provide traders with real-time insights.
Built and deployed a FastAPI-based chatbot on Azure AKS, handling 1,000+ daily queries with 99.8% uptime.
Automated chunking and indexing of 10K+ documents using Azure Form Recognizer and Cognitive Search via Azure Functions.
Implemented secure, scalable RAG system with JWT & LLMOPS standards.
Designed input cards and real-time CAQ, achieving 85%+ relevance in contextual responses and suggested queries.
Designed a scalable forecasting framework for utility sector use cases in collaboration with Microsoft, supporting stock prediction, demand forecasting, and power optimization.
Built and integrated 10+ FastAPI services for data ingestion, feature engineering, EDA, and forecast setup, streamlining the modeling workflow.
Automated CI/CD processes with Azure DevOps, reducing deployment effort by 70% through provisioning of Azure resources.
Collaborated on a React-based UI and integrated Azure ML, Blob Storage, Azure SQL, and PowerBI for end-to-end data flow and real-time insights.