I have a large basketball box scores statistics database. This contains data about individual players, teams, and events.
Currently, we have ~5 tables with all this information.
Im creating a model that can answer question like “Out of the top 10 scorers in the NBA this season, which players have the highest turnovers?”.
Our current workflow is: Text - > Foundational Model to generate SQL → run SQL on DB → FM to generate response.
We are experimenting w/ OpenAI Assistants/LangChain/chaining the models ourselves, however, it’s taking too long.
Goal is to create something similar to https://www.statmuse.com/ search functionality.
Can Julius support this search to make it fast (sub 5 seconds)?