Capstone: AI-Powered Parts Database Search

In this capstone project, I was part of a team that developed an AI-powered search tool for a parts database. The goal was to improve the efficiency of finding specific parts for engineers working at a major biomedical device company. At a high level, the tool used a rendering engine to generate images of 3D CAD models, evaluated them using a CLIP model, and then compared user queries to the parts database using a cosine similarity score to return the most relevant results.

Procedure Applied:

• Designed and implemented a user-friendly interface for the search tool.
• Utilized NLP techniques to process and understand user queries.
• Developed a recommendation system using collaborative filtering methods.

Skills Applied:

• Computer Vision(Marqo, CLIP)
• API integration (FreeCAD, Blender, Marqo)
• Server and application integration (Docker, http server-client communication)

Key Outcomes:

• Gained practical experience in applying AI techniques to real-world problems.
• Achieved a significant improvement in search efficiency and user satisfaction.

Photos cannot be shown due to NDA concerns