Fuzzing is a software testing technique used to find security, stability, and correctness issues. A program’s input domain is typically too large to enumerate exhaustively. Instead, we use e-graphs to improve input generation for fuzzing compilers. We implement this technique for fuzzing Wasm compilers, interpreters, and validators. Our approach, wasm-mutate, produces a sequence of Wasm programs that are semantically equivalent to its original given input. Wasm-mutate represents the search space for new programs as an e-graph and exploits the property that any traversal through the e-graph represents a semantically equivalent variant of the input program. This talk will focus on the proposed algorithm to traverse the e-graph to provide semantically equivalent code variants and the ways we leverage wasm-mutate while fuzzing.