Meta Machine Learning Engineer interview with a conversational-assistant design and declined offer

The process dragged on longer than I expected. It felt drawn out from the early weeks onward. I went through screenings on core ML knowledge, including attention and its variants, before later stages shifted into more high-pressure technical work.

Onsite day was intense and tightly structured. In the system-design round, I had to build a conversational assistant and include safety measures. That pushed me beyond model ideas into guardrails and real-world behavior. I also had AI-enabled coding interviews where using an AI assistant was not penalized. I used that kind of collaboration to reach correct solutions quickly, but I still felt I had to show more hands-on coding instead of leaning too heavily on the workflow.

I ultimately received an offer, but declined it because it did not feel like the right fit. What stayed with me was the demanding mix of deep ML concepts and system-level design in a process that lasted long enough for the pressure to build.

Location: United States. Overall feedback: neutral. Offer status: got offer.