Show HN: Reliopt – Pareto-frontier, contract-gated optimization for LLM programs

Category: ai-ml

Tags: llm, multi-objective-optimization, pareto-frontier, contract-testing, agent-framework

Score: 5.5/10 (Innovation: 6, Technical: 5, Documentation: 6, Utility: 5)

Reliopt is a Python library for multi-objective optimization of LLM programs, using hard contracts to gate candidates and returning Pareto frontiers rather than a single weighted score. It also provides component attribution via zero-ablation and automated stress-testing with perturbed inputs, making it a promising meta-layer for evaluating and selecting AI program architectures.

Target audience: ai engineers, backend developers, ml researchers

Repository: https://github.com/obielin/reliopt · Python · MIT · 1 stars

View on Hacker News