nt.number theory - Reference for Deligne's construction of Galois representations attached to modular forms

I absolutely agree that Deligne is very terse. One thing that I ultimately found very helpful is Carayol's two papers where he proves the analogous theorem for Hilbert modular forms. I say "ultimately" because it took me a long time to read those papers. I would come back to them every few years and learn more, as I matured mathematically. The big problems with using Carayol to understand Deligne will be: (1) Carayol has to work much harder in places than Deligne, because the Shimura curves he uses are not the solution to a moduli problem of abelian varieties plus extra structure, so he has to use extra tricks which Deligne did not have to get into, and this will obfuscate things (I guess perhaps this is only when analysing the bad reduction of the curves, which is perhaps not the paper you'd be wanting to read anyway) and (2) Deligne had to deal with the fact that modular curves need compactifying, so he had to work with parabolic cohomology, which is a technicality he has to deal with and Carayol doesn't. But for the main part, the techniques are the same.