InternshipProjects - Travel Assistant & Safety Guardrails

Complete collection of 7 internship projects (D1-D7) focused on:

• 🗺️ Travel Assistant APIs using Google Gemini
• 🔗 RAG (Retrieval Augmented Generation) systems
• 🧠 Memory-enhanced conversational AI
• 🛡️ Safety Guardrails & PII Detection

---

📁 Project Structure

InternshipProjects/
├── D1/ - Travel Assistant (Basic Gemini API)
├── D2/ - Travel Assistant (Prompt Engineering)
├── D3/ - Travel Assistant (LangGraph State Management)
├── D4/ - Travel RAG Assistant (Qdrant Vector DB)
├── D5/ - Memory-Enhanced Travel Assistant (Mem0 + Redis)
├── D6/ - Advanced Travel RAG (Caching + Monitoring)
├── D7/ - Safety Guardrails (PII Detection + Toxic Content)
├── TREE.md - Complete file structure
├── API_SAFETY_GUIDE.md - Important safety information
└── .gitignore - Git configuration (sensitive files excluded)

---

🔒 IMPORTANT: API SAFETY

This repository has been cleaned for GitHub upload with MAXIMUM safety:

✅ Deleted:

• ❌ All .env configuration files (API keys)
• ❌ All __pycache__/ Python cache
• ❌ All *.pyc compiled files
• ❌ All *.zip submission files
• ❌ All .cache/ directories
• ❌ All output.txt logs

✅ Preserved:

• ✅ All source code (.py files)
• ✅ All Jupyter notebooks (.ipynb)
• ✅ Documentation (README.md in each project)
• ✅ Dependency lists (requirements.txt)
• ✅ Project structure

🔐 NO API KEYS are stored in this repository.

---

🚀 Quick Start (Development Only)

1. Clone Repository

git clone https://github.com/YOUR_USERNAME/InternshipProjects.git
cd InternshipProjects

2. Choose a Project

cd D1/travel-assistant  # or any D2-D7

3. Create Environment File

cp .env.example .env  # If it exists
# Edit .env and add your API keys

4. Install Dependencies

pip install -r requirements.txt

5. Enable API (Optional - For Local Development)

Edit main.py and change:

API_DISABLED = True  # Change to False

6. Run Application

# For FastAPI projects
python main.py

# For Jupyter notebooks
jupyter notebook D*_assignment.ipynb

---

📚 Project Details

D1: Travel Assistant - Basic

• Goal: Learn Google Gemini API with FastAPI
• Key Concepts: LLM prompting, rate limiting
• Models: Gemini 2.5 Flash (FREE)
• Cost: $0 (within free tier)

D2: Travel Assistant - Prompt Engineering

• Goal: Master prompt templates and engineering
• Key Concepts: PromptTemplate, chain prompting
• Models: Gemini 2.5 Flash
• Cost: $0 (within free tier)

D3: Travel Assistant - LangGraph

• Goal: Implement state machines for agent workflows
• Key Concepts: State graphs, agent routing
• Models: Gemini 2.5 Flash
• Cost: $0 (within free tier)

D4: Travel RAG - Vector Database

• Goal: Build retrieval-augmented generation system
• Key Concepts: Embeddings, Qdrant, vector search
• Services: Gemini + Qdrant (open-source)
• Cost: $0 (open-source Qdrant)

D5: Memory-Enhanced Travel Assistant

• Goal: Add persistent memory to conversations
• Key Concepts: Mem0 service, Redis caching, user sessions
• Services: Gemini + Mem0 + Redis
• Cost: ⚠️ Mem0 may have costs

D6: Advanced Travel RAG

• Goal: Production-grade RAG with monitoring
• Key Concepts: Hybrid search, caching, metrics, tracing
• Services: Embeddings + Qdrant + Redis + Prometheus + Langfuse
• Cost: ⚠️ Langfuse may have costs

D7: Safety Guardrails

• Goal: Implement multi-layer safety filters
• Key Concepts: PII detection, toxicity filtering, prompt injection
• Services: Presidio + Guardrails Hub + Gemini
• Cost: $0 (all components free/open-source)

---

🛠️ Technology Stack

Core LLM

• Google Gemini 2.5 Flash - FREE tier, fast inference

Frameworks

• FastAPI - Web API framework
• LangChain - LLM integration
• LangGraph - Agent state management
• Guardrails Hub - Safety validation

Data & Storage

• Qdrant - Vector database (open-source)
• Redis - Caching and sessions
• Mem0 - Persistent memory service

Monitoring

• Prometheus - Metrics collection
• Langfuse - LLM trace monitoring

Security

• Presidio - PII detection (Microsoft)
• Pydantic - Data validation

---

📋 Requirements

Each project has its own requirements.txt:

Common:

• Python 3.8+
• FastAPI
• Pydantic
• LangChain
• Google-Generativeai

Optional (for advanced projects):

• Qdrant-client
• Redis
• Mem0
• Prometheus-client
• Langfuse

---

🔐 API Keys Needed (For Running)

⚠️ CRITICAL: Never commit API keys to Git!

# Create .env file (DO NOT COMMIT)
GOOGLE_API_KEY=your_key_here
MEM0_API_KEY=your_key_here  # Only for D5
QDRANT_API_KEY=your_key_here  # Only for D4, D6

Get Free API Keys:

• Google Gemini: https://ai.google.dev
• Qdrant: https://qdrant.io (self-hosted or cloud)
• Mem0: https://mem0.ai
• Presidio: Open-source (no key needed)

---

📖 Documentation

• API_SAFETY_GUIDE.md - ⭐ Read this first!
• TREE.md - Complete file structure
• Each D/README.md* - Project-specific documentation
• Jupyter notebooks - Code examples with explanations

---

🧪 Testing Without APIs

Run code without making actual API calls:

# Test imports
python -c "import langchain, fastapi; print('✅ OK')"

# Check project structure
python -c "import os; print([d for d in os.listdir('.') if d.startswith('D')])"

# View notebook structure
jupyter nbconvert --to script D1/travel-assistant/D1-assignment.ipynb --stdout | head -50

---

⚙️ Configuration

Environment Variables Template

Create .env file in project root:

# API Keys
GOOGLE_API_KEY=sk-...
MEM0_API_KEY=...
QDRANT_API_KEY=...

# Server
API_PORT=8001
API_HOST=0.0.0.0

# Database
REDIS_HOST=localhost
REDIS_PORT=6379
QDRANT_URL=http://localhost:6333

# Monitoring (Optional)
LANGFUSE_API_KEY=...
PROMETHEUS_PORT=8000

# Safety (Optional)
PRESIDIO_API_KEY=...

---

🔄 Development Workflow

1. Explore: Read Jupyter notebooks first
2. Understand: Review README.md in each project
3. Setup: Create .env with API keys
4. Run: Execute main.py or start API
5. Test: Use provided test_api.py files
6. Learn: Check comments in source code

---

🐛 Troubleshooting

ImportError: No module named 'langchain'

pip install -r requirements.txt

GOOGLE_API_KEY not configured

# Create .env file with real API key
echo "GOOGLE_API_KEY=your_key_here" > .env

Port already in use

# Change API_PORT in .env
API_PORT=8002

Rate limit errors

• Wait a few minutes (free tier has 10 req/min limit)
• Use delays between API calls

---

📜 License

This is an educational internship project repository.

---

✨ Features Demonstrated

✅ LLM API Integration (Google Gemini) ✅ FastAPI Web Framework ✅ Prompt Engineering & Templates ✅ State Machine Workflows (LangGraph) ✅ Vector Database Integration (Qdrant) ✅ Caching Strategies (Redis, Semantic Cache) ✅ Memory Services (Mem0) ✅ Safety & Guardrails (Presidio, Toxic Detection) ✅ Monitoring & Metrics (Prometheus, Langfuse) ✅ PII Detection & Masking ✅ Rate Limiting & Error Handling ✅ Jupyter Notebook Documentation

---

📞 Support

For issues or questions:

1. Check the README in specific project folder
2. Review API_SAFETY_GUIDE.md
3. Check API documentation for each service
4. Refer to Jupyter notebooks for examples

---

Status: ✅ Ready for GitHub Upload
Date: January 28, 2026
Safety Level: 🟢 MAXIMUM (APIs disabled by default)