requirements.txt

fastapi
# Web framework for building high-performance REST / async APIs
# Used for LEO Activation & CDP service endpoints

fastapi-pagination
# Pagination utilities for FastAPI, standardized limit/offset and cursor-based pagination

prometheus-fastapi-instrumentator
# Prometheus metrics integration for FastAPI
# Provides automatic metrics collection and endpoint for Prometheus scraping

uvicorn
# ASGI server to run FastAPI
# Handles async IO, WebSockets, and high concurrency

transformers
# Hugging Face library for LLMs and NLP models
# Used for local inference, embeddings, or model experimentation

torch
# PyTorch backend required by transformers
# Provides tensor computation and GPU acceleration

huggingface_hub
# Access models, tokenizers, and configs from Hugging Face Hub
# Handles authentication and model versioning

python-dotenv
# Loads environment variables from .env files
# Keeps secrets and configs out of source code

requests
# Simple HTTP client for calling external APIs
# Used for webhooks, third-party services, or legacy APIs

pydantic 
pydantic-settings
pydantic[email]
# Data validation and settings management
# Email utilities for user notifications and alerts

accelerate
# Hugging Face utility for efficient model loading and inference
# Helps with multi-GPU, mixed precision, and memory optimization

sentence-transformers
# Library for generating sentence and text embeddings
# Used for semantic search, clustering, and similarity tasks

google-genai
# Google Gemini SDK (latest)
# Used for LLM reasoning, function calling, and embeddings

openai
# OpenAI API client
# Used for GPT-4, GPT-3.5, embeddings, and other OpenAI services

psycopg[binary]
# PostgreSQL driver (psycopg v3)
# High-performance, async-friendly DB access

pgvector
# PostgreSQL extension client for vector embeddings
# Enables semantic search and similarity queries in Postgres

sqlalchemy
# ORM and SQL toolkit
# Used for schema management, migrations, and complex queries

pandas
# Data manipulation and analysis library
# Used for data processing, transformations, and analytics

plotly
# Interactive plotting library
# Used for visualizations in analytics dashboards and reports

python-arango
# ArangoDB driver
# Used for graph + document data (journeys, relationships, identity graphs)

celery[redis]
# Distributed task queue with Redis as broker
# Used for async jobs: embeddings, campaign execution, batch processing

redis
# In-memory data store
# Used for caching, rate limiting, Celery broker, and ephemeral state