A key point: have an agenda of what you want to discuss and share it with them. If you just ask to “pick their brain,” it comes off as you wanting the person to drop everything just so you can assail them with any question you can think of, potentially for hours. Check out his whole post for what makes the above example email effective.
If you have a good conversation with someone and they help you with something, follow up! For example, if you were looking for a job, let them know when you found one and thank them.
If you were introduced to someone or have a connection with them and haven’t heard back, give them a week or two and follow up. Sometimes emails just fall through the cracks and they’ll appreciate being reminded. If it’s a cold email, use your judgment about whether to follow up.
Finally, don’t take non-responses personally. People are busy, and you don’t know what is happening in their personal life.
Remember that others want to help; just look at how many have signed up to help answer questions and mentor data science newcomers at data helpers (this is a great resource for finding someone to reach out to). And the person who’s most helpful to you is probably not the most famous data scientist. You’ll likely learn more by talking with peers facing the same issues, teaching beginners to force a deeper understanding of core topics, or asking someone who recently accomplished your goal (e.g. giving your first talk).
The main way I try to help others in the community is through these posts, and I’ll be back to tackling the challenges of A/B Testing soon.
 Never talk to someone just to get an introduction to someone else; this goes back to the “not using people” principle.