Information Specialist / Volunteer Librarian — Greater Manchester Mental Health Library (AI training support via retrieval, verification, and feedback dataset structuring)
Designed information retrieval and verification workflows to support RLHF-style human feedback loops for RAG and LLM alignment use cases. Evaluated structural integrity, relevance, and semantic precision of text outputs to improve training signal quality and reduce hallucinations. Gathered and cleaned user feedback datasets to systematically analyze service quality and iterate dataset or interface inputs. • Conducted complex Boolean searches to find authoritative sources for grounding. • Synthesized high-fidelity information into structured training-ready bulletins. • Established quality control metrics and source authority evaluation standards. • Curated cleaned feedback datasets for iterative refinement cycles.