I maintain neat-python and have been building a genetic programming framework in Julia. While doing research on that project, I kept finding the same complaint in PySR's GitHub discussions: people give up not because symbolic regression doesn't work, but because getting Julia talking to Python is a repeated source of failure, especially on Windows and in notebooks.

So I stood up Occam (occam.fit) — a hosted MCP server exposing two tools: run_sindy for sparse identification of dynamics from time series (seconds), and run_pysr for evolutionary symbolic regression returning a Pareto front of expressions by complexity vs. accuracy (10–60s).

The core design decision is a pool of pre-warmed Julia workers running SymbolicRegression.jl v2. Julia processes stay alive between requests so the 2-minute compilation hit is paid once at server startup, not per call.

Free tier, no signup. Happy to answer questions about the architecture or symbolic regression generally, and curious whether anyone has use cases that don't fit the current limits.