Dylan Bristot is a product marketer working on Nebius Token Factory and a builder of AI research and model-selection tools. His work connects the infrastructure behind model deployment with explanations of what developers need to run agents in production.[1][2]
Start with Agent 101: Launching production-grade agents at scale →
Links
Website
X 
GitHub LinkedIn Founding Product Marketing Manager, AI Studio and Token Factory · Nebius
Career
Contributions
Explain the production stack around an agent
His Nebius guide separates the stack into models, agent frameworks, evaluation and memory. It discusses the operational problems that appear after a prototype works: cost visibility, silent failures, integration and changing traffic. The guide makes deployment an ongoing engineering task, rather than treating a successful model response as the end of the work. Models supply reasoning, frameworks coordinate tools, evaluations test behavior, and memory carries information between steps.[3]
Map constraints across the AI supply chain
AI Bottlenecks organizes research around the physical and commercial constraints on AI infrastructure. His other tools include WhatLLM and LLM Selector, which help users navigate model choices. These are product and research projects alongside his employer role.[1][4]
Tools, writing & talks
Writing, research & talks
Start here
A practical introduction to models, frameworks, evaluation and memory in a deployed agent system.
Read the four-part stack, then the sections on evaluating behavior and observing failures. Compare those requirements with a prototype you know: which costs, errors and changes in behavior would you actually notice after deployment?
Author[3]
Apps
- AI Bottlenecksaibottlenecks.app
A research interface for exploring constraints across the AI infrastructure supply chain.
Builder and publisher[1][4]Original source