AI Agent Architecture: LLMs, Tools, and the Hybrid Infrastructure Reality | Dr. Robert Blumofe, Akamai | TFiR
Akamai CTO Dr. Robert Blumofe explains why production AI agents require hybrid CPU, GPU, and storage infrastructure, ...
AI Infrastructure
Artificial intelligence and machine learning discussions with industry leaders and practitioners. Covering AI infrastructure, large language models, GenAI adoption, MLOps, vector databases, AI security, and enterprise AI strategy from executives building real-world AI systems.
Serverless Inference as a Service: Removing the AI Infrastructure Bottleneck | Jon Alexander, Akamai | TFiR
Jon Alexander of Akamai explains why single-region cloud architecture creates barriers to AI scaling, how to plan ...
Why AI Infrastructure POCs Stall Before Production and How to Fix It | Rob Hirschfeld, RackN | TFiR
Rob Hirschfeld, CEO at RackN, explains why hand-built AI clusters never reach production and how Digital ...
How to Decide Where AI Inference Should Run: Edge vs. Centralized | Ari Weil, Akamai | TFiR
Ari Weil of Akamai breaks down which AI inference workloads require edge distribution, which do not, and ...
Model-Agnostic AI Architecture and the Cost of Agent Interface Fragmentation | Mario Moscatiello, Airbyte | TFiR
Mario Moscatiello of Airbyte explains why MCP, SDK, and CLI interfaces will coexist in enterprises, why consistent ...
Why AI Inference Infrastructure Must Be Distributed, Not Centralized | Dr. Robert Blumofe, Akamai | TFiR
Dr. Robert Blumofe of Akamai explains why AI training and inference require completely different infrastructure architectures, why ...
What Enterprises Should Demand From AI Infrastructure Before Scaling Inference | Jon Alexander, Akamai | TFiR
Jon Alexander of Akamai outlines what enterprises must evaluate before large-scale AI inference deployments: portability, elasticity, storage ...
Edge AI Inference: The Last-Mile Problem Centralized Clouds Cannot Solve | Ari Weil, Akamai | TFiR
Ari Weil of Akamai explains why centralized GPU clusters cannot meet the latency demands of fraud detection, ...
Stop Applying Training-Era Hardware Logic to AI Inference Workloads | Rob Hirschfeld, RackN | TFiR
Rob Hirschfeld of RackN explains why context window management, not GPU compute, is the primary bottleneck in ...
How to Monitor and Control AI Agents Across Enterprise Infrastructure | Moshe Bar, Codenotary | TFiR
Moshe Bar of Codenotary explains how AgentMon and AgentX give enterprises immutable audit logging, karma-based agent scheduling, ...