This job is the perfect for me. I am the head of analytics at my company in Toronto, Canada. I normally use freelancer.com to get graphic designers to build virtual shelves for the conjoint work that my company does, but I saw this job and created my own personal account for it.
a) It depends on your data. If it's conjoint data, I would run latest class analysis to determine segments. Otherwise I would run k-means cluster analysis on attitudinal or behavioral data (possibly need states, if that's what you want). Looking for a 5 or 6 cluster solution, unless you had something else in mind. I would ipsatize (normalize) the data to avoid getting a "high" group and/or a "low" group (i.e., people usually will cluster according to how they used the scale as opposed to their actual similarities, so this trick prevents that).
b) I'm assuming you have primary data and not just a theoretical case study. I would use the conjoint data to derive utilities and then create a market simulator in Excel. This would allow us to run market shares for any new product we can imagine (based on the inputs available).
c) I would increase and decrease other product's price points and track how they affect Kirin's new product. Based on that, we would know which of its competitors are closest. We can also run a perceptual map to tell us how close Kirin's new products are to others to get a data visualization of cannibalization.
d) Easily done by looking at respondent level utilities.