📈 Multi-Agent Autonomous Equity Research Platform

An enterprise-grade, multi-agent financial intelligence framework that processes real-time market data and unstructured corporate disclosures into cross-verified equity research reports.

Powered by CrewAI, Llama-3.3-70B (via Groq), ChromaDB, AgentOps, and Streamlit.

🌟 Key Features

• Autonomous Multi-Agent Architecture: Leverages CrewAI for deterministic, role-based orchestration (Quantitative Market Analyst, Qualitative RAG Specialist, Compliance Lead).
  • Architectural Note: Benchmarked against LangGraph and AutoGen; CrewAI was selected for its rigid task delegation, which is critical for financial compliance and eliminating infinite loops.
• Enterprise RAG Pipeline: Built with ChromaDB and semantic embedding models using advanced knowledge engineering techniques (dynamic chunking and automated vector ingestion).
• Zero-Hallucination Guardrails:
  • Context-Leakage Prevention: Strictly grounds qualitative outputs, rejecting irrelevant RAG context when ticker symbols mismatch.
  • Symbol Pre-Validation: Pre-checks valid exchange tickers (US & NSE India .NS) before agent invocation.
  • Deterministic Fallbacks: Enforces max_iter=3 and explicit N/A return policies to prevent agent infinite loops.
• Observability & Cost Optimization:
  • AgentOps Integration: Live session tracing, token tracking, and agent execution performance monitoring.
  • API Cost Control: Optimized inference routing using Groq's high-throughput llama-3.3-70b and local vector search caching.
• Streamlit Dashboard: Features dynamic disclosure uploads, real-time status rendering, and instant .md dossier export.

🏗️ Architecture Overview

                        +----------------------------+
                        |   Streamlit Frontend UI    |
                        +--------------+-------------+
                                       |
                   +-------------------+-------------------+
                   |                                       |
         [ Stock Ticker Input ]                 [ File Ingestion ]
                   |                                       |
                   v                                       v
      +------------------------+              +------------------------+
      |  CrewAI Execution Engine|             | Save to ./data & Run   |
      |  (Monitored by AgentOps)|             |      ingest.py         |
      +-----------+------------+              +-----------+------------+
                  |                                       |
        +---------+---------+                             v
        |                   |                  +----------------------+
        v                   v                  | ChromaDB Vector Store|
  +-----------+       +-----------+            +----------------------+
  | yfinance  |       | ChromaDB  |                       ^
  | API Tool  |       | RAG Tool  |                       |
  +-----------+       +-----------+                       |
        |                   |                             |
        +---------+---------+-----------------------------+
                  |
                  v
       +-----------------------+
       | Compliance Lead Agent |
       +-----------+-----------+
                   |
                   v
       [ Verified .md Dossier ]

🛠️ Tech Stack & Frameworks

• Orchestration: CrewAI
• LLM Engine: llama-3.3-70b-versatile via Groq Cloud API
• Vector DB & RAG: ChromaDB, Sentence-Transformers
• Observability: AgentOps
• Market Data: yfinance
• Frontend UI: Streamlit
• Language: Python 3.11+

📁 Repository Structure

EQUITY-RESEARCH-PLATFORM/
├── data/                  # Raw corporate filings and disclosures (.txt/.pdf)
├── src/
│   ├── agents.py          # Agent definitions, roles, and LLM configs
│   ├── crew.py            # CrewAI workflow orchestration
│   ├── tasks.py           # Task descriptions and compliance guardrails
│   └── tools.py           # Custom yfinance and ChromaDB search tools
├── vector_db/             # Local ChromaDB persistent storage
├── .env                   # API Key configuration
├── app.py                 # Streamlit web frontend
├── ingest.py              # Document chunking & embedding script
└── requirements.txt       # Dependencies

🚀 Getting Started

1. Prerequisites

Ensure you have Python 3.10+ installed, along with API keys for Groq and AgentOps.

2. Installation

Clone the repository and install required packages:

git clone https://github.com/your-username/equity-research-platform.git
cd equity-research-platform
python -m venv venv
source venv/bin/activate  # On Windows: venv\Scripts\activate
pip install -r requirements.txt

3. Environment Setup

Create a .env file in the root directory:

GROQ_API_KEY=your_groq_api_key_here
AGENTOPS_API_KEY=your_agentops_api_key_here

4. Running the Application

Launch the Streamlit web app:

streamlit run app.py

💡 Usage

1. Enter Ticker: Provide a valid US ticker (e.g., NVDA, AAPL) or NSE ticker (e.g., RELIANCE.NS, RPOWER.NS) in the sidebar.
2. Ingest Documents (Optional): Upload custom corporate filings or pdfs through the sidebar widget to dynamically build the local RAG knowledge base.
3. Execute Analysis: Click Run Multi-Agent Squad to initiate the workflow. Monitor live agent execution on your AgentOps Dashboard.
4. Export Report: Download the finalized Markdown dossier using the built-in export widget.