12Dinesh/stocksense-mlops — reverse-engineered prompt

Build me a complete StockSense project that predicts whether popular stocks like AAPL, MSFT, NVDA, and GOOGL are likely to move up or down. It should pull real data from Yahoo Finance, save it locally, create technical indicator features, train LightGBM and XGBoost models, tune them, and track experiments with MLflow.

I want a simple dashboard where I can choose a ticker, see recent prices, view the model prediction, confidence, useful charts, and basic monitoring like drift or data quality warnings. Also create a FastAPI service so predictions can be requested through an API, plus a simple setup that can run locally with Docker Compose for things like Redis, Prometheus, and Grafana if needed.

Please make it easy to run from a fresh checkout with clear commands, sensible defaults in config, and a quick test pipeline so I can confirm everything works. Look up current docs online if you need to.