How are actions of agents debugged in multi-agent systems?
Machine Learning / AI
Tell about a support request classification task for prioritization.
How to rank extracted chunks by relevance to the query?
Tell about a successful classic model — employee churn prediction.
How are agent metrics and logs sent to Grafana (code)?
How to correctly group data by salary categories without duplicate rows?
Do you have questions about the technical part?
Calculate the average work experience for each salary category
Redis is non-persistent — what will happen if it restarts? Will we lose notifications?
How to highlight and visualize agent metrics in Grafana?
What is the idea of a random forest and why does it work?
How to find the last purchase by a person in the data?
How did you collect metrics for the quality of the RAG system? What metrics did you use and what tools?
What is LangGraph? Why use it instead of LangChain or raw API?
What is LoRA?
How does the LLM extract the necessary data from the request to pass to the tool?
What methods of text chunking do you know?
What types of memory exist in LangGraph? How are short-term and long-term memory implemented?
How do you evaluate the relevance of extracted chunks (accuracy of extraction)?
What components would you use to build a system for searching documents and information about people (locally, with GPU)?