Title

Student Research Chatbot – Helping students get faster, grounded answers

Short outcome: Built an AI assistant that answers academic questions using a curated document set, as a way to practice designing and iterating an AI-powered research tool.

Context

• What problem existed

  Students and early researchers spent a lot of time jumping between papers, websites, and notes to answer targeted questions. Search tools returned links, not tailored explanations, and generic chatbots lacked the right context.

• Who the users were

  Students and early-stage researchers who needed quick, focused answers from academic material without reading every source end to end.

• Why it was worth solving

  • Slow research workflows reduce learning efficiency and delay insight generation.
  • Many student questions are specific but repeatable, which makes them a good fit for a focused assistant.
  • This was a good space to practice real AI product decisions around model choice, retrieval quality, and user experience.

My Role

• Ownership and responsibilities
  • Defined the problem based on how students currently search and read for answers.
  • Translated needs into requirements for retrieval, response quality, and interaction flow.
  • Scoped and prioritized the MVP, including what to measure and what to leave for later.
• Scope and constraints
  • Self-driven learning project, not a production system.
  • Limited compute and tooling, so model and architecture choices needed to be pragmatic.
  • Focused on building a usable prototype that could support iteration, not a polished, large-scale product.

Problem Statement

Students struggled to get concise, trustworthy answers to research questions from multiple academic sources.

Search engines and library tools provided documents, not explanations. Generic chatbots could answer in fluent language but often lacked grounding in the student’s actual materials.

There was a need for an assistant that could answer research questions based on a specific, curated document set, not the general web.

Goals & Success Metrics

• Goals
  • Help students get quicker, more relevant answers to research questions.
  • Ensure answers are grounded in the right sources rather than generic model knowledge.
  • Learn how to design and iterate an AI product beyond a one-off demo.