Introducing Vector Embeddings and Vector Search in the jobdata API
jobdataapi.com v4.13 / API version 1.15
UPDATE: We released an update and provide all embeddings as 768-dimensional half-precision floats - please read the full changelog here
We are introducing vector embeddings and vector search capabilities to the jobdata API, enabling advanced semantic analysis and search functionalities for job postings. These features leverage OpenAI's text-embedding-3-small model to generate embeddings for every job post, available as 1536-dimensional embeddings (embed_3s) for detailed semantic representation. By activating these embeddings in API responses, you can perform deeper analysis, such as clustering job posts by semantic content or matching job listings with candidate profiles based on contextual similarity.
Our new (experimental) vector search functionality allows you to perform semantic searches on job listings using the vec_text parameter. When a text query (up to 1000 characters) is provided, the API converts it into embeddings in real-time and compares it against job embeddings using cosine similarity. This enables a more meaningful and context-aware matching of job listings, even when exact keywords are not present in the job description.
These new features are available now with the new access pro+ and our existing access ultra subscription plans. To put them into action, include the embed_3s or vec_text parameters in your API requests on the /api/jobs/ endpoint. Detailed documentation and examples are available in the Vector Embeddings and Search API Documentation.