ROBINSON OJWANG

Executed large-scale LLM evaluation and annotation programs, built prompt libraries and meta-prompts, and owned data augmentation and QC pipelines to boost evaluator throughput and dataset diversity.

Invented cognitive narrative scoring and shipped prompt-to-brand-voice controls and safety filters, improving editorial acceptance by 17% and reducing human rewrite time by 28% while cutting policy violations by 41%.

Built classification and relevance pipelines for tone, formality, and contextual rewrites, enabling style suggestions with ~9.6% adoption and improving inter-rater agreement for preference datasets.

Contributed to commonsense reasoning benchmarks, authored prompt suites for abductive reasoning, and implemented data collection and adjudication pipelines to strengthen evaluation sets.

Led projects on factuality metrics for dialogue models and taught workshops on annotation labeling, rater calibration, and content understanding for digital humanities projects.

Delivered multilingual labeling and relevance tasks at production scale, designed gold standards and QC tiers, and improved pass rates and escalation processes for ambiguous cases.

Prototyped similarity evaluation and data augmentation for lesson generation, improving distractor selection and producing A/B learning gains of +6.1%.

Implemented discourse parsers and dashboards for newsroom editors to detect bias and assess content quality as part of computational journalism projects.

Activities and societies: Dissertation research; development of evaluation harnesses; workshops and collaborations with Adobe on narrative metrics.

PhD program in Creative Writing & Literature focused on AI and literary cognition, including a dissertation on cognitive models for AI-generated narrative and development of evaluation frameworks for narrative quality.

Activities and societies: Graduate research, annotation workshops, corpus construction, and teaching annotation/rater calibration.

Master of Arts in Linguistics with a Computational Linguistics focus and a Certificate in Digital Humanities; thesis on measuring factuality and coherence in generative dialogue systems and work on multilingual corpora and annotation schemata.

Activities and societies: KnightLab research assistant, speech & debate, outreach in digital literacy.

Bachelor of Science in Computer Science and Communication Studies with concentrations in linguistics, logic and symbolic reasoning, and NLP fundamentals; capstone on media bias detection via discourse parsing.