Problem
DevGlobe already supports semantic/vector search, but discovering peers still requires the visitor to invent a search query. Developers do not have a one-click way to find people with a similar technical profile, which limits the product's value for networking and collaboration.
Proposed solution
Add a Developers like me experience for a signed-in or selected developer. Return nearby developer profiles from the existing vector index and explain the strongest visible similarities.
Acceptance criteria
Technical notes
app/api/search/route.js already uses Cosmos DB VectorDistance; add a profile-to-profile query that accepts a developer identity rather than embedding a free-text query.
- The embedding text currently includes profile, location, language, and repository signals. Similarity should be described as profile similarity, not a quality score or rank.
- Avoid presenting a precise percentage unless the distance is calibrated against the dataset; a label such as
Very similar is safer initially.
Related issues
Problem
DevGlobe already supports semantic/vector search, but discovering peers still requires the visitor to invent a search query. Developers do not have a one-click way to find people with a similar technical profile, which limits the product's value for networking and collaboration.
Proposed solution
Add a Developers like me experience for a signed-in or selected developer. Return nearby developer profiles from the existing vector index and explain the strongest visible similarities.
Acceptance criteria
Developers like meaction on the authenticated user's profile and aSimilar developersaction on any developer detail viewTechnical notes
app/api/search/route.jsalready uses Cosmos DBVectorDistance; add a profile-to-profile query that accepts a developer identity rather than embedding a free-text query.Very similaris safer initially.Related issues