Using the Amazing World
summary
Joblift builds a pipeline that moves from attribute boosting in Elasticsearch to vector models: Doc2Vec on job descriptions to map CVs or freeform “dream job” text into similar postings, and item2vec from click sequences that treats job IDs as words to learn behavioral similarity. A live demo shows CV-to-job and description-to-job matches, but models are not in production: paragraph embeddings take ~30 seconds per request, training needs large compute and data, and major challenges are cold start, expired listings and noisy input.
AI-written summary · about
chapters (9)
- 0:00Introduction
- 1:24Requirements
- 4:49Boost Attributes
- 7:21First Step
- 8:03Knowledge Language
- 14:19Similarities
- 18:58Neural Networks
- 23:11Production
- 25:21Questions
related talks
18m: Workforce Efficiency with AI-Driven Vector Search
50mVector Ops: run vector embedding-powered apps in prod
31mTaking AI/ML to the next level: Text!
42mDocument Vectors in the Wild
45mA gentle introduction to Vector Databases
27mMachine Learning for Advanced AI Search
39mDeveloping Machine Learning for Impact
31mAdvanced search for your e-commerce website