Work

Chatbot Knowledge Expansion

Human-centric design
Data curation
Natural Language Processing

Training AI chatbots before it was cool.

Robot head with machines at each side pulling it apart. Made with AI.

Partner Operations Intern | AI Edu-Tech Startup

Summary

In a pre-ChatGPT era, I spearheaded efforts to refine a chatbot’s knowledge base, transforming it from a chaotic patchwork of outdated info into a scalable, student-centric tool. Amid a team incentivized for speed over quality, I identified systemic flaws—like redundant answers inflating rankings—and initiated a cleanup: removing outdated content, streamlining responses, and ensuring every answer prioritized student needs.

Beyond data, I explored the human side of AI: impersonating the chatbot to engage students, escalating unmet needs like financial stress, and highlighting the importance of empathy in design. This experience underscored that AI success hinges not just on technical precision, but on aligning workflows, incentives, and human expectations.

The project shaped my focus on sustainable AI practices, leading me to advocate for data curation and onboarding new universities with clean, intentional knowledge bases.

Impact
  • Scaled AI chatbot capabilities by curating and updating its knowledge base over 750 times weekly, ensuring accuracy and relevance for thousands of student interactions.
  • Expanded access to resources by researching and integrating materials from over 100 partner universities and organizations, enhancing the chatbot’s utility for academic and administrative support.
  • Enabled seamless onboarding for new partner schools, implementing chatbots that reached thousands of students, bridging gaps in access to critical university resources.
Tech Stack
  • Proprietary AI platform and NLP model

Read the full story here on my Substack.