Portfolio Optimizer

This project won 1st Place at the Cloudera Evolve24 Hackathon.

For the Cloudera Evolve24 Hackathon, this project focuses on optimizing Large Language Models (LLMs) through Retrieval-Augmented Generation (RAG) using a Pinecone vector database. The goal is to gather financial news headline data and, based on this data, recommend actions to perform on a simulated trading account via the Alpaca API (in paper trading mode). By combining user information with vectorized headline data, processed through Pinecone, and integrating it into our own deployed model on Nvidia's NIM platform, the system provides tailored financial recommendations for enhanced decision-making.

Demo and video:

Demo link: https://315c591179f68ee8fd.gradio.live/ The demo is no longer live after the duration of the hackathon.

Video: https://youtu.be/Uzz8NhnQHXE

Table of Contents

• Features
• Prerequisites
• Installation
• License
• Cloudera Workbench

Features

• Grabs Latest Headline Data using Perplexity AI.
• Vectorization Storing with Pinecone.
• Alpaca trading simulation with Alpaca API.
• ChatGPT API integration for recommendations.

Cloudera Workbench

• This project utilized the Cloudera CML workbench, jobs for raw data ingestion, sessions for easy coding and optimizations, AMPs for usecases, and it's app deployment to host our application

Prerequisites

• Language/Platform version (Python 3.8 or higher)
• Accounts and API keys for:
  • OpenAI
    • CHATGPT_API_KEY
  • AlpacaAPI
    • ALPACA_API_KEY
    • ALPACA_SECRET_KEY
  • PerplexityAPI
    • PERPLEXITY_YOUR_API_KEY
  • NvidiaAPI
    • NVIDIA_API_KEY
  • PineconeAPI
  • PINECONE_API_KEY
  • FMP_API_KEY

Installation

Clone the Repository

git clone https://github.com/jjl9895/PortfolioOptimizer.git
cd PortfolioOptimizer

License

This project is licensed under the MIT License.