Always make the project plan in collaboration with the people who will be doing the work.
Ask probing questions to understand the risks. Is there technical risk? Is there product definition risk? Is there a risk for workscope creep?
Take the time and care to understand the approaches and estimates presented by developers, and make sure assumptions are aligned. Sometimes they aren’t – and there is a better and faster way just under the surface, if you ask good questions.
Consult people who have done similar types of projects for a sanity check: “Are we crazy?”
Special thanks: Martin Trust Centre, MIT
Edit: Javier Rojas, 15/04/20