HN Showcase
Back to browse
GitHub Repository

Vibing a more performant ocaml sexp parser than Jane Street

2 starsOCaml

I vibed a better OCaml parser than Jane Street in 69 steps

by yevbar·Feb 20, 2026·3 points·1 comment
Visit ProjectView on HN

AI Analysis

●●●BangerWizardryBig Brain

AI agents beat Jane Street's parser; Phase 3 won all benchmarks with 37% less code.

Strengths
  • •Genuine performance win: measurable, runnable locally, 7/7 benchmarks won; not hand-waved claims.
  • •Meta-insight: swarm self-improvement (constraint + test generation) beat single-agent; reproducible methodology.
  • •Correct execution: Phase 3 traded speed on edge cases for 37% smaller code and test coverage explosion.
Weaknesses
  • •Niche codebase: OCaml + sexp parsing is specialist domain; won't influence mainstream tooling.
  • •Proof-of-concept framing: 'vibing' suggests playfulness; doesn't address Jane Street's maintenance burden or production requirements.
Target Audience

OCaml developers, compiler engineers, performance-critical systems

Similar To

Cranelift (Rust codegen, performance-focused alternative) · Tree-sitter (parser generator that beats hand-written)

Post Description

*for some cases

Using cloud sandboxes to run them in I tested: - A single coding agent just told to make a better parser - An agent told to write a better parser within the constraints of tests/benchmarks - An agent swarm that self-improved the premise with extra tests/benchmarks in order to more "truly" write a better parser

The results were a success! I was able to end up with both performance (up to 3.07× faster) and memory (up to 5.75× less) in locally runnable benchmarks.

Similar Projects