What would you like to do in general?
Middle
155
Machine Learning / AI
3611
0
What can be done when the agent's context gradually fills up during long work?
Senior
152
How to store large information received via API in the state in LangGraph, use it inside a tool, but not pass unnecessary data into the LLM context? For example, the LLM only sees IDs and names of specialists, while the tool gets their schedule and additional data.
Senior
150
How can the quality of a production monitoring/incident investigation system be evaluated?
Senior
149
147
If you need to train an LLM not to answer when there is no information in the context, how to do it?
Senior
146
How to practically make the model generate only values from a given short list of string IDs and no other characters in a specific field?
Senior
146
What can be done if a deployed agent responds too slowly or lacks memory for hosting LLM?
Senior
144
142
Who was involved in tuning the models?
Middle
141
How to pass a schema for structured output to the model at the code level?
Senior
141
Tell me more about KV-cache: how it works and where it comes from, starting from the Transformer architecture.
Senior
140
How did you implement and deploy a service for processing pull requests?
Senior
139
Does the orchestrator see the results of sub-agent tool calls or only the final output of the sub-agent itself?
Senior
138
What other optimization methods can be used to speed up inference or at least fit the model into 12 GB VRAM RTX 3090?
Senior
135
How to verify that LLM-as-a-Judge correctly assesses quality and how to calibrate it?
Senior
133
How is the orchestrator built: what framework and architecture are used, what does it do, is it a one-step or can it reason multiple times before answering?
Senior
132
How to achieve a formalized, structured response, for example when using structured output?
Senior
132
What are the options for sampling new tokens in LLM?
Senior
131
What is your current team composition?
Middle
125