benoitc/erlang-python

Provides multiple parallelism strategies (worker processes, free-threaded Python 3.13+, and sub-interpreters with isolated or shared GIL) to match performance needs, with process-bound Python environments enabling OTP-style supervised Python actors. Supports bidirectional calls between Python and Erlang/Elixir, async/await patterns, generator streaming, and shared state via thread-safe ETS-backed counters. Designed for AI/ML workloads with built-in examples for embeddings and RAG, plus security sandboxing to prevent VM corruption from fork/exec operations.