Goal
Design a minimal system that uses an LLM-based AI Agent to collect missing details from a user (User Info, Vehicle Info, ZIP) and then return a price quote once the required data is complete.
You are not expected to build a full pricing engine. Use a simple deterministic pricing stub.
Part 1 — Agent Prompt The agent must:
* Accept partial input in any order.
* Ask only for missing required fields.
* Validate basic formats (ZIP = 5 digits; year reasonable; mileage numeric).
* Produce strict JSON output every turn (no extra text).
Required fields to collect:
* userInfo: fullName (required), contact (required)
* vehicleInfo: year, make, model, trim (optional), mileage (required), condition (required: one of excellent|good|fair|poor)
* zipCode (required)
When all required fields are present and valid, return status="READY_FOR_PRICING" with the normalized payload. When you have all the details you can generate a random price
Otherwise return status="NEED_MORE_INFO" and include nextQuestion.
Part 2 — REST API Design + Pseudo Code
Design a RESTful service that:
1. Receives user messages / partial input.
2. Calls the agent with session context.
3. Either returns the agent’s next question OR returns a final price quote.
Interview questions [1]
Question 1
You said:
Goal
Design a minimal system that uses an LLM-based AI Agent to collect missing details from a user (User Info, Vehicle Info, ZIP) and then return a price quote once the required data is complete.
You are not expected to build a full pricing engine. Use a simple deterministic pricing stub.
Part 1 — Agent Prompt The agent must:
* Accept partial input in any order.
* Ask only for missing required fields.
* Validate basic formats (ZIP = 5 digits; year reasonable; mileage numeric).
* Produce strict JSON output every turn (no extra text).
Required fields to collect:
* userInfo: fullName (required), contact (required)
* vehicleInfo: year, make, model, trim (optional), mileage (required), condition (required: one of excellent|good|fair|poor)
* zipCode (required)
When all required fields are present and valid, return status="READY_FOR_PRICING" with the normalized payload. When you have all the details you can generate a random price
Otherwise return status="NEED_MORE_INFO" and include nextQuestion.
Part 2 — REST API Design + Pseudo Code
Design a RESTful service that:
1. Receives user messages / partial input.
2. Calls the agent with session context.
3. Either returns the agent’s next question OR returns a final price quote.
Raj will ask you screenshare and be demanding
I applied online. The process took 2 weeks. I interviewed at Copart (Dallas, TX) in Jan 2024
Interview
They asked multiple Q regarding Python(theory based) followed by Few SQL Question and at the end concluded by asking about Unix, OS, networking. but i wasnt given any feedback nor result and was ghosted at end
Smooth and great, everything was laid out perfectly. Had first call with the HR to match me to the job description. After that got a technical interview with one of the senior devs. After that it was a HR interview with the Technical Director and the higher management.