← Back to homeongoingongoingongoing
Research & Experiments
Technical exploration and learning — from agent planning to voice and RAG.
Adversarial LLM Evaluation System
A multi-model reasoning framework that reduces hallucinated claims through structured disagreement and hybrid judging logic.
RAG evaluation experiments
Benchmarking retrieval quality and answer faithfulness for product-knowledge RAG systems.
Algorithmic Exposure Simulation Lab
A synthetic audience engine that models how content distribution emerges from interaction dynamics rather than direct prediction.