In the Table 1.12 of the book Aroon V. Shenoy, Rheology of Filled Polymer Systems, Kluwer Academic Publishers, Netherlands (1999), there are a number of suggested surface modifiers that would work for the epoxy resin matrix – A-187, A-1100, A-1120, CG6720, CA0750, CA0700, GLYMO, AMEO, DAMO-P, Z-6040, Z-6020, KR-TTS, KR38S, KR-46B, KR-138S, LICA 44. These would be from the categories of silanes, and titanates and their full forms are given in Table 1.8. Which one would be most effective for Al2O3 nanoparticles can only be worked out by some mixing experiments, which are discussed in the book. Particularly for Alumina, one could think of the zircoaluminate (CAVCO MOD APG) surface modifier as it might turn out to be very effective.
Investigation of this publication may be useful for you:
Ghorabi, Shima, Laleh Rajabi, Sayed Siavash Madaeni, Sirus Zinadini, and Ali Ashraf Derakhshan. "Effects of three surfactant types of anionic, cationic and non-ionic on tensile properties and fracture surface morphology of epoxy/MWCNT nanocomposites." Iranian Polymer Journal 21, no. 2 (2012): 121-130.
Article Effects of three surfactant types of anionic, cationic and n...
Anionic surfactants such as SDS and SDBS seem to be good for this purpose. These emulsifiers interact with the alumina nanoparticles, and prevent their agglomeration.
To reach epoxy resin, generally no solvent is used during the hardening process. Epoxy resin and curing agent should be mixed with each other. I think you must homogeneously disperse your alumina nanoparticles within the resin (such as DGEBA), and then add the hardener to the reaction vessel.