Traditional meta-compilation frameworks—most notably PyPy’s RPython—achieve execution speed by translating high-level dynamic languages at the bytecode level into intermediate control-flow graphs for whole-program type inference. While powerful, this approach carries substantial architectural complexity and runtime tracing infrastructure. This paper introduces minipy, a lightweight, high-performance Python interpreter built using an extended variant of RPython. Moving away from traditional bytecode-level translation, minipy employs a novel AST-driven flattening architecture integrated directly into the ShivyCX/py2c compilation pipeline. By emitting direct, statically optimized C components that integrate bare-metal compiler passes—such as register-resident global flag packing, non-recursive static buffer relocation, and tail-call elimination—the resulting interpreter bypasses typical virtual machine overhead. Empirical evaluations show that minipy significantly minimizes execution footprints and outperforms CPython in critical micro-benchmarks.