I've looked into doing something similar to this before and love the idea, never had the chance to follow through on it yet. This is sort of a hybrid method between cast molding and 3D printing.
The accepted answer to a question I had a while back had some very good points by fred_dot_u
Post processing FDM for strengrh
Short version, Epoxy is a good option but you may have to consider heat generated from it. Urethane is another really good option (cast urethane is a pretty standard process). Chem-Eng isn't my area of expertise but there is a huge range of material options out there that can be mixed as two parts. I think there are enough options out there right now that you can choose your material properties you need and then select the material from there.
You could also consider going the chopped-fiber composite route. (carbon, glass, etc) and then combine with whatever the appropriate resin is for those materials.
I see the most difficult part of this is getting the shells to print properly. When I had looked into doing this, I considered modeling my part then hollowing it out completely. Then going back into the hollow part and designing in minimal internal structures for the purpose of supporting the thin-walled shell model. Printing that, and then drilling and filling the part after the fact. This approach I see as being a good option however the location of the drill points would be critical otherwise you could get voids as your fill material is injected in. And, the additional modeling time wouldn't be insignificant, however the saving I expected would come from having a ridiculously strong part, with complex geometry and be significantly cheaper than even a cast-urethane part.
If you get some good results, please post a link to them! This is a huge interest of mine!