Abstract: Dark matter dominated dwarf galaxies are excellent targets to constrain thermal dark matter annihilation, due to their clean background in gamma-ray regime. However, the common practice in the community for estimating the J-factor, a key component in calculating the strength of the annihilation signal, may be prone to systematic uncertainties. In this talk, I will discuss a new method for inferring dwarf galaxy J-factors from semi-analytic modeling and explore the associated systematics. In particular, we find qualitative differences in the J-factor inference between ultra-faint and classical dwarf galaxies. I will also discuss J-factor estimation for a number of recently discovered ultra-faint stellar clusters, including Ursa Major III. Lastly, we apply our inferred J-factor to the analysis of gamma-ray data from Fermi LAT telescope and present adjusted constraints on the dark matter annihilation cross section.