Google

PART 1: ALGORITHMIC PROBLEM - CONFERENCE ROOM ALLOCATION BASE PROBLEM You are given an array of meeting time intervals where intervals[i] = [start_i, end_i], representing the start and end times of a meeting (start_i < end_i). Task: Write an algorithm to determine the minimum number of conference rooms required to hold all the meetings without any overlapping conflicts in the same room. (Note: A meeting ending at time 't' and another starting at time 't' do not conflict). FOLLOW-UP 1 Instead of just finding the minimum number of rooms, the facilities team needs the actual schedule. Task: Modify your algorithm to return a valid assignment of meetings to specific rooms. The output should group the original intervals by the room they are assigned to (e.g., Room 1 gets meetings A and C, Room 2 gets meeting B). What data structures are best suited for this? FOLLOW-UP 2 The problem gets more complex. Now, each conference room has a maximum capacity 'C', and each meeting has a number of attendees 'A'. A meeting can only be assigned to a room if the room is available AND the room's capacity is greater than or equal to the meeting's attendees (C >= A). Task: Discuss how you would approach this 2D constraint problem (time + capacity). If it's not possible to accommodate all meetings, how would you modify the system to prioritize meetings with the largest number of attendees? PART 2: AI DISCUSSION & TERMINOLOGY Practical Application & Ethics: If you were tasked with building an AI system to screen resumes for a hiring manager, what are the potential risks of bias, and how would you try to mitigate them? General Knowledge: What is the difference between a standard conversational AI (like a basic chatbot) and an "AI Agent"? Basic Terminologies (Briefly explain the following concepts): Context Window: The maximum amount of text (measured in tokens) that an AI model can take in, remember, and process at one single time. If a conversation exceeds this limit, the model "forgets" the earliest parts of the chat. Zero-Shot Prompting: Asking an AI model to perform a task without giving it any prior examples or demonstrations in the prompt. (Conversely, "few-shot" is when you provide a couple of examples first). Inference: The phase where a trained AI model is actually being used to generate predictions or answers based on new user input. This is distinct from the "training" phase. Epoch: In machine learning training, one complete pass of the entire training dataset through the algorithm. Training usually takes multiple epochs for the model to learn effectively.