WEEK3
Embeddings & Vector Search
Semantic search over documents by meaning.
Course Details
Students learn how modern AI search systems understand information using embeddings instead of keywords. They explore document chunking strategies, vector databases, semantic retrieval, and build complete ingestion pipelines that transform raw documents into searchable knowledge. By the end of the week, students can create AI applications that retrieve information based on meaning rather than exact text matches.
Topics Covered
Embeddings & cosine similarity
Chunking strategies & their tradeoffs
Vector databases & indexing
Semantic retrieval by meaning
Building an ingestion pipeline
Topics Covered
OpenAI Embeddings
Pinecone / Qdrant
pgvector
Python
Project 3
KnowledgeVault · PaperFinder
Project Details
Project Results
Project Result 1
Developed a document intelligence application that converts unstructured documents into a searchable knowledge base using embeddings, chunking, vector indexing, and semantic retrieval. Implemented an end-to-end ingestion pipeline for efficient AI-powered document search.
Project Result 2
Created a semantic research paper search application using embeddings, vector databases, and cosine similarity. Built a complete ingestion, indexing, and retrieval pipeline that finds relevant papers based on meaning rather than keyword matching.
