M6
Vector Search
Embeddings + cosine similarity
Vector embeddings and approximate nearest neighbor search
Available
Cohort
- Effort
- 5-7 hours
- Prerequisite
- M5
- Core concept
- Semantic similarity
What you have
Lexical search only
What you gain
Semantic recall over the same documents
What you build
The module is planned. It will add a second retrieval path that complements the lexical system built earlier.
- An embedding pipeline that converts product text into vectors
- A vector index for nearest-neighbor retrieval
- A retrieval path that returns semantic candidates for the same corpus
What you learn
- How lexical retrieval and semantic retrieval answer different failure modes
- What vector similarity adds when exact terms are missing
- Why recall gains often introduce new ranking trade-offs
Artifact and workload
Primary artifact: Embedding pipeline and vector index
Tests25
Assessments5
Estimated time5-7 hours
Access
This module is part of the cohort. Join the guided path for reviews, deadlines, and the workshop sequence after the ranking modules.