AI Product Consulting · User Research · Workflow Design · Knowledge Systems · Rollout & Training
Two distinct AI product engagements across social impact sectors: • Project 1: AI Teaching Assistant for Media Literacy (Education) - Co-designed an AI-powered teaching assistant for children's media literacy programs, transforming social topics and curriculum scripts into a structured, reusable knowledge base. - Developed prompt frameworks and multi-turn dialogue flows aligned with different learner personas, including differences in age, comprehension level, and engagement style. - Evaluated platform feasibility across low-code AI tools and designed access-control strategies for controlled pilot rollout. - Focused on improving pedagogical consistency, response reliability, and scalability of content delivery. • Project 2: AI Support Bot for Elderly Services (Senior Care) - Conducted stakeholder interviews to assess operational workflows such as approval, finance, and data handoff, and identified system inefficiencies and unmet user needs. - Designed workflow logic and system architecture for a support bot integrating text knowledge, video content, and web search. - Processed and structured knowledge assets using Python to improve tagging, retrieval, and response accuracy. - Led usability testing and onboarding, optimizing interaction design for elderly users with lower digital literacy. - Supported deployment through training, permission configuration, and scenario-based testing in real-world settings.
• AI product design starts with human understanding: successful systems depend on aligning user capability, context, and cognitive load, not just model performance. • Capability-aware design matters: designing for elderly users versus young learners showed that usability changes dramatically with digital literacy, decision behavior, and interaction patterns. • Effective AI products require systems thinking: strong solutions come from integrating knowledge base design, prompt logic, workflow mapping, and access control into one coherent experience. • Reusable knowledge systems outperform one-off setups: modularizing content into structured knowledge bases significantly improves scalability, consistency, and long-term maintainability. • Real product consulting is an end-to-end loop: stakeholder interviews → workflow mapping → solution design → testing → training → iteration. Adoption depends as much on alignment and enablement as on technology. • The hardest part is not building the bot, but making it understandable, trustworthy, and usable for different groups in real-world contexts. • Through training and co-building (“vibe coding”), I also observed how individuals differ in learning speed, abstraction ability, and execution, which deepened my interest in capability-based evaluation beyond resumes.