VIC - Virtual Institutional Counselor

Context

Designed around the reality that academic requirements, policies, and procedural information are usually spread across different documents and sources.

Problem

Students need quick answers about academic processes, but the source material is fragmented, repetitive, and difficult to navigate consistently.

Solution

VIC uses Python, Streamlit, PostgreSQL, NLP, and ETL pipelines to transform institutional content into a searchable guidance workflow that is easier to query and maintain.

Key decisions

• -Used Streamlit to move quickly on an interface centered on guided information retrieval rather than custom frontend complexity.
• -Added an ETL layer so institutional data could be normalized before retrieval instead of relying on raw source text.
• -Kept PostgreSQL as the structured storage layer to support repeatable lookups and rule-driven responses.

Key features

• -Streamlit web UI for guidance sessions
• -NLP layer to classify and route requests
• -ETL pipeline to ingest and normalize institutional data
• -PostgreSQL storage for structured program and rule data
• -Reproducible local setup for demos and iteration

Results

• -Turned institutional guidance into a workflow that is faster to consult and easier to demonstrate.
• -Connected applied AI concepts to a concrete academic use case instead of a generic chatbot wrapper.
• -Created a stronger portfolio example of data preparation, retrieval logic, and user-facing delivery working together.

Stack

Links