We would like to build an web-based chat interface for users that will reply with a forecast of financial stock prices as well as a chart with this forecast and other, known data. We have historical stock prices - along with other relevant documents/articles - that we would like to give the OpenAI model access to (via vector store/RAG/etc) for improving the model's forecasts/responses.
Roughly once per day, we would like to update the data (vector store/RAG/etc), so the model has the most up-to-date information.
First, we will go over a consultation that covers our goals and the steps you recommend to implement in order to accomplish these goals. Here is a possible option:
1. Store historical data prices and documents in an OpenAI vector store
2. Create an assistant using the OpenAI Assistants API, passing this vector store in the constructor
3. Deploy a Flask server that connects to the Assistants API and responds to users' queries.
- this Flask server/Assistant may generate a chart using Function calling, and calling a separate API endpoint
4. Set up a cron job to regularly update the vector store/RAG/etc.
This job must be complete by November 11.
Following this initial project, we may choose to expand the scope in order to improve the model’s forecasting performance, again based on your suggestions, for example integrating a financial forecasting model.
Location: Anywhere
Posted: Oct. 30, 2024, 4:24 a.m.
Apply Now Company Website