[presentation]

Late but - often what I do is, if a task is big enough that I don't feel I can really estimate it, I decline scheduling that task in a sprint, and instead schedule a research task to explicitly spend time scoping/speccing out a complex task. During that time I try to make smaller, stepping stone tasks that I think I can reasonably estimate. I intentionally pad those a bit, and also say that I gave some buffer to account for unknown unknowns, which is a reasonable thing to do.

Also, asking other people on your team to talk it through with you is a good idea as well - they can help you see some unknowns you might have missed that would have blown up your estimates, and also give their take on effort required. If you're uncomfortable with saying you want to work with someone because you don't feel confident in designing a solid solution, you can still couch it in other terms, like that you want to bounce ideas off of someone else on the team, parallelize your work by delegating subtasks, or make sure others on your team are familiar with the design and implementation - all legitimate reasons to work with others, but also that deflect a feeling of just "not knowing".