At Americana Restaurants, I built and deployed Amira, a natural-language analytics agent over Databricks that supports about 500 queries a day. Prompt caching reduced its API cost by about 70%, and the agent was piloted with 20–30 users ahead of a planned wider rollout.
I also built GenAI advertising products using Vertex AI, FastAPI, Streamlit, and pgvector RAG, saving the marketing team days per cycle and more than $10,000 annually. My work includes data pipelines and optimization systems, including a transport solver that cut fleet size by 10% and transport cost by more than 20%.
