If you mentor anyone for longer than a few months, the subject will drift. You start out talking about testing strategy and pull requests, and one Tuesday they ask whether they should take a job in another city, or how to handle a manager who talks over them, or whether the sixty hour weeks they are doing are normal. That drift is a good sign. It means they trust you. It is also the point where a lot of mentors either freeze or overreach, and both are worth avoiding. Freezing looks like steering back to the technical topic because the other thing is not your job. It reads as a door closing, and they will not knock twice. Overreaching looks like answering everything with total confidence. You are not their manager, their lawyer, their doctor, or their family, and some of what they bring you needs one of those. Telling someone to quit a job when you know one side of the story is a heavy thing to be wrong about. The middle is simpler than it sounds. Listen properly first, without arranging your answer while they talk. Then be explicit about which hat you are wearing. I can tell you what I have seen happen to two people who took that kind of role, and what it cost them. I cannot tell you whether your immigration status survives a resignation, and you should not take my guess on it, so let us find someone who actually knows. Say what is experience and say what is opinion, and label the difference out loud. Offer the shape of a decision rather than the decision. What would make this obviously right. What would make it obviously wrong. What would you regret in a year. Two more things. Keep what they tell you completely to yourself, including from people who would find it interesting, and especially if you work anywhere near their employer. Trust dies in one sentence. And notice your own bias. You are recommending the path you took, or the one you wish you had taken, and neither of those is their life. You are not there to run their decisions. You are there so they do not have to make them alone. – Asael Shinder