Sure. To start with there are many, many models for personality profiling. I tend to stick to "The Big 5" mostly because it's foibles don't sufficiently affect the outcomes I care about, and I've gotten proficient enough at it that with the right set of questions in the right environment I can typically get a pretty good read on at least 4 out of the 5 factors, and given a follow-up interaction I can usually clear up anything that looks like contradicting signals by getting additional context.

Like for instance, if you were to meet me, you would probably immediately peg me as high on extroversion. While understandable given the outward traits I project, you'd have needed to really thoroughly created a controlled environment during an interview (requiring a series panel, a day-long engagement, and an intentional highly-social event in the middle) to figure out that I'm actually introverted, but have just learned how to benefit myself by projecting extroverted traits.

Now the richness of any assessment is bound by time, planning, training, and practice. If you want to get really good at it, you learn to do it in realtime by assessing people you know well, assessing yourself, then moving on to acquaintances, and eventually onto strangers. In each step you need to be aware of where your assessments are drifting from the model. Like if you assess yourself or someone else as high on agreeableness because you find that you're very typically conciliatory or trend toward diplomacy and keeping the peace amongst your peers, but you also are prone to "blow-ups" where you drop the hammer on people in rare, but consistent, circumstances (when you've "been pushed too far"), then you're probably not as high on agreeableness as you think, and are probably higher on neuroticism than you think you are.

Given enough inputs and hypotheses you can get pretty good at this. Good enough to provide useful points for making decisions given otherwise limited information. More importantly you can get good at how you setup your environments and your interactions to get higher quality signals from your limited information. For instance, say I wanted to assess someone's "Openness" because I wanted to bring them onto a team where their primary responsibility will be in driving phases of invention. The simplest pass would be to just ask them about personal activities and look for signals of risk-taking. However, one needs to control for the difference between risk-taking due to openness to experience and risk-taking due to self-destructive traits. The former would be awesome for the job role and the latter will bring down the whole ship. To control for that you'd want to put additional emphasis on assessing neuroticism and conscientiousness. Doing that can be as simple as leaving a lot of awkward pauses in the conversation and see how they deal with it. Does it make them uncomfortable? Do they feel compelled to fill the air with something? What do they fill it with? Relatable stories, trivia about their work history, etc. When they talk about anything, not just work-related things (sometimes specifically not work-related things), what level of detail do they provide? Is it consistent? Is it always TMI, always 30k ft view, or can they move fluidly from one to the other?

Things like that. Different kinds of job roles require a different kind of personality profile to be successful, and more important than that, the composite of personality profiles filling those job roles is a requirement for a healthy team where each person is competitive with themselves, but cooperative with their peers, and has the right mix of people to own each phase of the development and business process.