We model the non-thermal transient Swift J1644+57 as resulting from a relativistic jet powered by the accretion of a tidally-disrupted star onto a super-massive black hole .
Accompanying synchrotron radio emission is produced by the shock interaction between the jet and the dense circumnuclear medium , similar to a gamma-ray burst afterglow .
An open mystery , however , is the origin of the late-time radio re-brightening , which occurred well after the peak of the jetted X-ray emission .
Here , we systematically explore several proposed explanations for this behaviour by means of multi-dimensional hydrodynamic simulations coupled to a self-consistent radiative transfer calculation of the synchrotron emission .
Our main conclusion is that the radio afterglow of Swift J1644+57 is not naturally explained by a jet with a one-dimensional top-hat angular structure .
However , a more complex angular structure comprised of an ultra-relativistic core ( Lorentz factor \Gamma \sim 10 ) surrounded by a slower ( \Gamma \sim 2 ) sheath provides a reasonable fit to the data .
Such a geometry could result from the radial structure of the super-Eddington accretion flow or as the result of jet precession .
The total kinetic energy of the ejecta that we infer of \sim few 10 ^ { 53 } erg requires a highly efficient jet launching mechanism .
Our jet model providing the best fit to the light curve of the on-axis event Swift J1644+57 is used to predict the radio light curves for off-axis viewing angles .
Implications for the presence of relativistic jets from TDEs detected via their thermal disk emission , as well as the prospects for detecting orphan TDE afterglows with upcoming wide-field radio surveys and resolving the jet structure with long baseline interferometry , are discussed .