Hi everyone,

I recently cleared Round 1 for a Microsoft SDE-2 role and HR informed me that my Round 2 will be an LLD (Low-Level Design) interview where access to an LLM will be available (e.g., GPT, Claude, DeepSeek, etc.).

I'm a bit confused about what exactly this means and how it changes the expectations for the interview.

Some questions I have:

• Has anyone recently gone through a similar Microsoft round?
• How was the LLM expected to be used during the interview?
• Were you allowed to ask the LLM for code, design suggestions, API definitions, class structures, or design patterns?
• Was the interviewer evaluating the quality of prompts and how effectively you used the LLM, or was it still primarily a traditional LLD round?
• Did the interview involve designing AI-powered systems, or was the LLM simply a productivity tool available during the discussion?

I'm also curious about how one can effectively incorporate an LLM into an LLD interview without becoming overly dependent on it.

My current assumption is that the interviewer may be less interested in memorization and more interested in:

• Problem decomposition
• Design trade-offs
• Requirements gathering
• Ability to validate or critique LLM-generated solutions
• Engineering judgment

If anyone has recently experienced this format, I'd greatly appreciate hearing about:

1. The type of LLD question asked.
2. How the LLM was used during the interview.
3. What the interviewer seemed to be evaluating.
4. Any preparation tips or pitfalls to avoid.

Thanks in advance!