Sorry if I misuse the flair, idk if I should ask here or any other subreddit.
I'm a junior CV engineer, about 10 months into my first role at a consulting company. I've been asked to assess a list of computer vision features for a new project, categorizing each one as "feasible", "maybe feasible", or "not possible". The project is mainly a guided instruction AR application, with some CV validation to make sure the client performs each step correctly.
The problem is: I asked for time to do preliminary R&D and, at a minimum, build small PoCs for the uncertain ones so the assessment has some empirical basis. The answer I got was essentially, make an educated guess and give your reasoning by tommorow :).
I understand timelines are tight, but this doesn't feel right to me. CV features have so many hidden variables, lighting, occlusion, real-world variance, and hardware constraints, that something that looks straightforward on paper can completely fall apart in deployment. An educated guess without any validation feels like setting the project (and myself) up to make promises we can't keep.
My questions:
Is this kind of "guess-based feasibility" common, especially in smaller teams or early-stage projects?
As a junior, how do you handle being asked to commit to technical estimates you don't have enough data to back up confidently?
Is there a professional way to push back, or at least protect yourself when the estimate turns out to be wrong?
I'm not trying to be difficult, I genuinely want to give a useful answer. I just don't want to confidently say "yes we can do X" and have it blow up 3 months into development.
Any advice from people who've been here would really help. Feeling a bit lost on how to navigate this.