AssistX AI Engineer Test Case: Vacation Planner

This project implements a proof-of-concept AI vacation planner that can understand natural-language travel requests, interpret user preferences, check (mock) calendar availability, search for mock flights and hotels, and assemble a structured, day-by-day VacationPlan using an LLM-powered agent equipped with carefully defined tools. It also provides a dedicated booking endpoint that allows users to confirm a proposed itinerary once they explicitly approve it. The implementation combines FastAPI for the backend API, Pydantic v2 for data modeling and validation, PydanticAI for agent orchestration, OpenAI’s GPT-5-Nano for natural-language reasoning, and a lightweight in-memory storage layer to keep the proof-of-concept fully self-contained and easy to run.

Click here to learn more about the project: assistx-vp/assets/Task - AI Engineer (LLM) Revised.pdf.

📁 Project Structure

assistx-vp
│
├─ app/                              # Solution app
│  ├─ config.py                      # Configuration files
│  ├─ main.py                        # FastAPI app
│  │
│  ├─ agent/
│  │  └─ vacation_agent.py           # PydanticAI agent with tools
│  │
│  ├─ models/
│  │  ├─ api.py                      # API models
│  │  └─ domain.py                   # Domain models
│  │
│  ├─ services/
│  │  ├─ bookings.py                 # Booking service
│  │  ├─ calendar.py                 # Calendar service
│  │  ├─ preferences.py              # Preferences service
│  │  ├─ sessions.py                 # Session helper
│  │  └─ travel_search.py            # Travel search service (for mock flights/hotels)
│  │
│  └─ storage/
│     └─ in_memory.py                # In-memory storage
│
├─ assets/
│  ├─ vacation_planner_solution.pdf  # Solution report
│  └─ vacation_planner_demo.gif      # Solution demo video
│
├─ .env
└─ requirements.txt

💡 Solution Report and Solution Demo Video

• The solution report includes overview, solution, and vulnerability and risk. Click here to learn more about the solution report: assistx-vp/assets/vacation_planner_solution.pdf.
• The solution demo video shows the working app, accessible via api call. Click here to learn more about the solution demo video: assistx-vp/assets/vacation_planner_demo.gif.

🔌 API

1. Health Check

   • GET /health: to verify that the backend is running correctly, responding to requests, and using the expected LLM configuration
     • Request: None
     • Response: 'status', 'model'

     curl "http://localhost:8000/health"

2. User Preferences

   • GET /api/preferences/{user_id}: to retrieve the user’s saved travel preferences or automatically initialize defaults if none exist
     • Request: user_id
     • Response: PreferencesResponse

     curl "http://localhost:8000/api/preferences/{user_id}"

   • PUT /api/preferences/{user_id}: to update the user’s travel preferences with new budgets, origins, interests, or other settings
     • Request: PreferencesUpdateRequest
     • Response: PreferencesResponse

     curl -X PUT "http://localhost:8000/api/preferences/{user_id}" \
     -H "Content-Type: application/json" \
     -d '{
        "home_city": "{home_city}",
        "default_currency": "{default_currency}",
        "max_budget_total": {max_budget_total},
        "max_budget_per_day": {max_budget_per_day},
        "interests": ["{interest_1}", "{interest_i}", "{interest_n}"],
        "travel_style": "{travel_style}",
        "preferred_airlines": ["{preferred_airline_1}", "{preferred_airline_i}", "{preferred_airline_n}"],
        "preferred_hotel_types": ["{preferred_hotel_type_1}", "{preferred_hotel_type_i}", "{preferred_hotel_type_n}"]
     }'

3. Chat w/ LLM

   • POST /api/chat: to send a natural-language message to the AI vacation-planning agent and receive a reply, potentially including a generated VacationPlan
     • Request: ChatRequest
     • Response: ChatResponse

     curl -X POST "http://localhost:8000/api/chat" \
     -H "Content-Type: application/json" \
     -d '{
        "session_id": "{session_id}",
        "user_id": "{user_id}",
        "message": "{message_to_llm}",
        "allow_booking": {bool}
     }'

4. Book Plan

   • POST /api/book: to confirm and record a booking for the latest AI-generated vacation plan within the user’s session, using a provided payment token
     • Request: BookRequest
     • Response: BookResponse

     curl -X POST "http://localhost:8000/api/book" \
     -H "Content-Type: application/json" \
     -d '{
        "session_id": "{session_id}",
        "user_id": "{user_id}",
        "payment_token": "{payment_token}"
     }'

⚙️ Local Setup

0. Make sure to have the prerequisites:

   • Git
   • Git Large File Storage
   • Python
   • Conda or venv

1. Clone the repository:

    git clone https://github.com/verneylmavt/assistx-vp.git
    cd assistx-vp

2. Create environment and install dependencies:

   conda create -n assistx-vp python=3.11 -y
   conda activate assistx-vp
   
   pip install -r requirements.txt

3. Fill the required OPENAI_API_KEY in .env

4. Run the server:

   uvicorn app.main:app --reload

5. Open the API documentation to make an API call and interact with the app:

   start "http://127.0.0.1:8000/docs"