Meet Rimantas Zukaitis, a Lead System Architect at EIS Group with over 20 years of experience in business rules, model-driven development, and low-code tools.
At PyCon Wroclaw 2026, Rimantas will help us look beyond the usual AI optimisation tricks in his talk:
🔍 “LLM Context Profiling: what is eating up your tokens?”
AI coding agents can consume 20 to 50 times more input tokens than the output they generate. But what exactly fills their context windows? And how can we reduce token usage without sacrificing precision?
Rimantas will introduce a hands-on approach to understanding and profiling LLM context with a custom-built tool. Think of it as memory or CPU profiling, but for AI context! You’ll learn how to spot unnecessary, duplicated, and low-value information before it drives up costs, increases latency, or affects model performance.
At PyCon Wroclaw 2026, Rimantas will help us look beyond the usual AI optimisation tricks in his talk:
🔍 “LLM Context Profiling: what is eating up your tokens?”
AI coding agents can consume 20 to 50 times more input tokens than the output they generate. But what exactly fills their context windows? And how can we reduce token usage without sacrificing precision?
Rimantas will introduce a hands-on approach to understanding and profiling LLM context with a custom-built tool. Think of it as memory or CPU profiling, but for AI context! You’ll learn how to spot unnecessary, duplicated, and low-value information before it drives up costs, increases latency, or affects model performance.