Exact probabilistic inference is a requirement for many applications of probabilistic programming languages (PPLs), but implementing a PPL with high-performance inference is difficult. Roulette is a new discrete PPL that combines high-performance exact inference with expressive language features by leveraging the close connection between exact probabilistic inference and the symbolic evaluation strategy of Rosette. Building on this connection, Roulette generalizes and extends the Rosette solver-aided programming system to reason about probabilistic rather than symbolic quantities. In this talk, I’ll demonstrate how to use Roulette, discuss some of the ideas that make it work, and describe ongoing work to extend it further.