.env.example

# =====================================================
# RAG Capstone Project - Environment Configuration
# =====================================================

# =====================================================
# LLM PROVIDER SELECTION
# =====================================================
# Choose your LLM provider: "groq" (cloud) or "ollama" (local)
# Ollama provides unlimited local inference with no rate limits!
LLM_PROVIDER=groq

# =====================================================
# OPTION 1: GROQ API (Cloud - Rate Limited)
# =====================================================
# Single API key
GROQ_API_KEY=your_groq_api_key_here

# Multiple API Keys for Rotation (optional - comma separated)
# Create multiple free Groq accounts to get more API keys
# This allows automatic rotation when rate limits are hit
GROQ_API_KEYS=key1_here,key2_here,key3_here

# =====================================================
# OPTION 7: OLLAMA (Local - Unlimited)
# =====================================================
# For unlimited local inference without rate limits!
# Install Ollama: https://ollama.ai/download
# Pull models: ollama pull gemma3:12b && ollama pull llama3.3
# Start server: ollama serve

OLLAMA_HOST=http://localhost:11434
OLLAMA_MODEL=gemma3:12b
OLLAMA_MODELS=gemma3:12b,llama3.3

# =====================================================
# VECTOR STORE CONFIGURATION
# =====================================================
# Choose vector store: "chroma" (local) or "qdrant" (cloud/persistent)
VECTOR_STORE_PROVIDER=chroma

# ChromaDB Configuration (local storage - ephemeral on HuggingFace Spaces)
CHROMA_PERSIST_DIRECTORY=./chroma_db

# =====================================================
# QDRANT CLOUD (Persistent Storage for HuggingFace Spaces)
# =====================================================
# Sign up free at: https://cloud.qdrant.io
# 1. Create a free cluster
# 2. Copy the URL and API key
QDRANT_URL=https://your-cluster.aws.cloud.qdrant.io:6333
QDRANT_API_KEY=your_qdrant_api_key_here

# =====================================================
# Google Gemini API Configuration (for gemini-embedding-001)
GEMINI_API_KEY=your_gemini_api_key_here

# =====================================================
# RATE LIMITING (Groq API only)
# =====================================================
GROQ_RPM_LIMIT=30
RATE_LIMIT_DELAY=2.5

# Retry settings for rate limit errors
MAX_RETRIES=3
RETRY_DELAY=60.0

# Batch processing with checkpoints
BATCH_SIZE=10
CHECKPOINT_ENABLED=true

# Application Configuration
LOG_LEVEL=INFO