AI / ML Internship Assignment
Objective
Scenario
Task Instructions
Step 1: Understand The User Context
Step 2: Build Your Prompt System
Step 3: Deliverables
Bonus Challenge (Optional)
Evaluation Criteria
Estimated Time
Objective
Building an AI agent system to guide travelers with their travel plans/inquiries and finally help
them generate a personalized travel itinerary
Scenario
You are building an AI-powered travel planner that provides users with highly personalized
travel itineraries. Your task is to prototype an AI-agentic system for this use case and write
appropriate prompts that can simulate a conversation to gather user preferences, use a
web-search tool to curate and suggest relevant items, and finally generate a detailed travel
itinerary. Host your application on any free hosting provider like streamlit, gradio, etc for user
testing.
Task Instructions
Step 1: Understand The User Context
1. Go through the user messages for inputs such as:
a. Budget
b. Trip Duration or Travel Dates
c. Destination & Starting Location
d. Purpose
e. Preferences
2. Your goal is to:
a. Extract key details from the user.
b. Chain prompts effectively to refine inputs if needed.
c. Generate a coherent, day-by-day itinerary.
Step 2: Build Your Prompt System
1. Input Refinement: Create a system prompt to gather or clarify additional details from
the user, such as:
a. Dietary preferences.
b. Specific interests within the given preferences.
c. Walking tolerance or mobility concerns.
d. Accommodation preferences (luxury, budget, central location, etc.).
2. Activity Suggestions: Use web-search tools, with chained prompts if required, to:
a. Generate accurate up-to-date suggestions for top attractions or activities at
the provided destination locations.
b. Align and contextualize these suggestions with user preferences (e.g.,
“Hidden Gems”).
3. Output: Finally once the suggestions are approved and the user requests a detailed
itinerary, design the final prompt that generates a complete, well-structured n-day
itinerary, with a logical grouping and timing of sightseeing and other activities for
each day
Step 3: Deliverables
● Submit the final set of prompts you used. Label each as System Prompt, User
Prompt, and Model Response.
● Submit some sample inputs and outputs:
○ Some curated suggestions for the provided user inputs
○ A detailed day-by-day itinerary for the provided user scenario.
● Document your process in a short paragraph for each prompt:
● Host the application on any free hosting service such as streamlit, gradio, etc, and
submit the link to test the live application.
Bonus Challenge (Optional)
Improve the itinerary by allowing for flexible input formats. For example:
● If the user says, “I have a moderate budget and I want a mix of famous and offbeat
places,” refine it to generate precise suggestions.
● Handle vague or incomplete inputs and prompt the user to clarify them.
Evaluation Criteria
➢ Prompt Design
○ Are the prompts clear, specific, and structured to guide the model effectively?
○ Do they ensure relevant outputs?
➢ Prompt Chaining
○ Are you using multiple prompts strategically to refine responses?
○ Do the prompts maintain a coherent flow?
➢ Personalization
○ How well does the output align with the user’s preferences?
○ Are the results unique and well-suited for the user scenario?
➢ Documentation
○ Are you explaining your reasoning and thought process effectively?
Estimated Time
You should ideally spend not more than the suggested time on each activity of the
assignment.
● Prompt Design & Testing: 3-4 hours
● AI agent design: 3-4 hours
● Hosting: 1-2 hours
● Documentation: <1 hours
