At Microsoft AI Skills, I developed a local RAG assistant with Python and Streamlit that reduced average information lookup time by approximately 80% in measured tasks. I added token-aware chunking and hybrid retrieval, with page-level references to help users verify answers against source documents.
I also validated the assistant’s PDF processing, retrieval, and application workflows with automated tests. This work gave me hands-on experience building and testing RAG applications.
At MedArt Hair Clinic, I developed and maintained responsive web solutions using Node.js and JavaScript, and integrated website content into WordPress so it could be updated through the CMS. I also applied SEO practices to support the clinic’s online presence.
For my CogniTwin capstone project, I co-developed a multi-agent platform with four role-specific workspaces. I implemented ontology-based grounding, shared memory, and bounded revision loops, and containerized the platform with Docker.
