At Mastercard, I developed an MCP server that uses LLMs to translate business questions into SQL queries over Mastercard transaction data. I also created a reusable causal inference package, with computationally intensive components implemented in C++.
I led methodology updates for Mastercard SpendingPulse country-level products in the UK and Australia, and designed automated ETL pipelines with monitoring dashboards and alerts.
At Moody’s, I worked on LLM-powered RAG applications and led quantitative analytics delivery. I re-engineered a financial statement scoring pipeline, cutting compute cost by about 78% and average execution time by at least 50%.

