GenAI Retail Forecast

GenAI Retail Forecast is an end-to-end MLOps pipeline that predicts demand across 1,115 store-product combinations and turns the numbers into an executive-readable narrative instead of a spreadsheet. Apache Beam and PySpark handle feature engineering — rolling lags, calendar signals, seasonal decomposition — feeding per-store ARIMA_PLUS models in BigQuery ML, while a RAG narrative layer grounds the generated summaries in the actual retrieved forecast data.