Automated Analyst Report Pipeline

A five-component autonomous pipeline that ingests supply chain data, retrieves relevant financial commentary via vector search, generates a professional analyst report using Llama 3.1 70B, and emails it as a formatted PDF — on a schedule, with zero human input.

Architecture

Query → Parser (Groq) → SQL Retrieval (SQLite) → Vector Retrieval (FAISS) → Context Assembly → Report Generation (Groq) → PDF (ReportLab) → Email

Sample Output

See outputs/reports/sample_report.pdf for a real generated report.

Key findings from the supply chain dataset:

• Haircare: highest defect rate (2.48%), Cosmetics: lowest (1.92%)
• 36 SKUs failed inspection; SKU42 at 4.94% defect rate
• Carrier C: best revenue-to-cost ratio ($1,138.56 per $1 shipped)
• SKU34: critical stockout risk — 1 unit in stock, 26-day lead time

Setup

Prerequisites

• Python 3.10+
• Groq API key (free): https://console.groq.com
• Kaggle account + API token: https://www.kaggle.com
• Gmail account with App Password enabled

Installation

git clone https://github.com/YOUR_USERNAME/analyst-report-pipeline cd analyst-report-pipeline python -m venv venv source venv/bin/activate # Windows: venv\Scripts\activate pip install -r requirements.txt

Configuration

cp .env.example .env

Edit .env and add your GROQ_API_KEY and EMAIL_PASSWORD

Data

Download the datasets via Kaggle CLI:

kaggle datasets download -d harshsingh2209/supply-chain-analysis
-p data/structured --unzip

kaggle datasets download -d ankurzing/sentiment-analysis-for-financial-news
-p data/unstructured --unzip

Run

Open analyst_pipeline.ipynb in Jupyter and run all cells in order. The scheduler in the final cell runs the pipeline automatically.

Stack

Component	Technology
LLM	Groq / Llama 3.1 70B
Vector Search	FAISS + Sentence Transformers
Structured Data	SQLite + Pandas
PDF Generation	ReportLab
Scheduling	APScheduler
Email	Gmail SMTP