Amono AI: A Parameter-Efficient Pluralistic Alignment Framework for Mitigating Western Monoculture Defaultism in Large Language Models

State-of-the-art Large Language Models (LLMs) deployed across global contexts exhibit pronounced normative monoculture, routinely prioritizing Western, Educated, Industri alized, Rich, and Democratic (WEIRD) ethical paradigms while marginalizing non-Western epistemologies. This structural bias imposes unilateral normative baselines on pluriversal societies. We present Amono AI, an open-source, prompt-conditioned framework architected on Google AI Studio and the Gemini API that enforces epistemic pluriversality through a multi-framework deliberative mechanism termed the Council of Epistemic Minds. Rather than arbitrating a single moral verdict, Amono AI dynam ically projects ethical dilemmas across four parallel philosophical traditions: (1) Indic / Dharmic Ethics, (2) Collectivist / Communal Ethics, (3) Indigenous & Biocentric Stewardship, and (4) Western Liberal Frameworks under strict, mode-dependent token budgets. We validate the framework across a two-tier empirical audit: a 5-scenario micro-benchmark on gemini-3.7-flash (N = 10) achieving 100% word-budget compliance, and an extended 50 dilemma macro-scale benchmark across five societal domains on gemini-3.6-flash (N = 100). Across all 100 macro-scale trials, Amono AI maintained 92.0% Compact compliance (≤ 100 words) and 100.0% Analytic compliance (≤ 250 words), achieving a maximal Shannon Equitability score of EH = 1.0000. Our results substantiate that structured prompt conditioning provides a computationally lightweight, robust mechanism to neutralize monocultural alignment bias without catastrophic forgetting or parameter retraining.