At PwC India, I built a Power BI MCP-based proof of concept to automate DAX measure extraction, repointing, and validation without XMLA endpoints or Power BI Premium. I also developed Python scripts to identify broken measures against a new star schema and programmatically repoint dashboard visuals.
I evaluated Databricks Genie's multi-agent architecture for root cause analysis, auto-documentation, and DAX validation, then proposed a partitioned Multi-Genie Space and Claude Opus 4.8 orchestration approach. I also benchmarked GPT-5.5, Gemini 3.1Pro, and Claude Opus 4.8 for enterprise root cause analysis and produced research and a stakeholder presentation adopted by the advisory team.
In my projects, I built a stock insights dashboard for portfolio monitoring and a machine-learning pipeline to predict SpaceX Falcon 9 landing success. I also developed and deployed a credit card fraud detection app, using resampling to address class imbalance and balancing precision and recall.
