Computing and recommending company-wide employee training pair decisions at scale...
summary
An in-house system called MAGNETRON AI automates mentor–mentee matching using R, Python and LLMs with a Shiny front end, MySQL and Neo4j back end and packages like reticulate, bs4dash, DT, plumber, dbplyr and neo4r. The engine filters by cohort eligibility and asymmetric compatibility, then sorts by averaged asymmetric ranks (skill similarity and activity distance). It shrinks manual matching from months to days, handles 84,000 potential pairs, and already matches a 400‑member data science community and larger mentoring cohorts for thousands, demonstrating practical success.
AI-written summary · about
chapters (13)
- 0:00Employee training matching intro
- 1:03Regeneron's challenge
- 3:52Manual pairing fails at scale
- 4:36Doctor Strange inspiration
- 5:40Magnetron AI for matching
- 7:52Decision science framework
- 9:38Framework for therapeutics
- 12:09Magnetron AI outcomes
- 13:13Recommendation engine
- 16:28Magnetron AI with R/Python
- 17:35Shiny app workflow
- 19:26Learn more
- 20:08Closing & Q&A
related talks
25mEnhancing Job Matching With Knowledge Graphs and RAG
42mDataEngBytes 2023 - SYD-T1-02 - Mikiko Bazeley
24mEmpowering Lifelong Learning: Transforming Data into Insights With GraphRAG
19mHacking Productivity with LLMs
22mOvercoming Loneliness with LLM Dating Assistant
13mOptimizing Real-Time Job Matching: Integrating Neo4j With RAG
01h 36mTutorial: From Planni...
15mGraphRAG-Powered AI Agents: Transforming Corporate Metadata into Intelligent Knowledge