Frameworks for LLM Agents have been surprisingly disappointing. I want to offer some principles for building agents based on our own trial & error, and explain why some tempting ideas are actually quite bad in practice.

Collection of awesome LLM apps with AI Agents and RAG using OpenAI, Anthropic, Gemini and opensource models. - Shubhamsaboo/awesome-llm-apps

Share your videos with friends, family, and the world

Contribute to cfregly/ai-performance-engineering development by creating an account on GitHub.

Is bigger always better? How our optimization space is fundamentally changing. The slow death of scaling and what comes next.

The recording of my talk on the slow death of scaling given at Khipu is now available.

This captures many of the ideas I'm currently…

Why traditional PMF playbooks are failing in 2025, and the new framework that's working (AI PRD Template included)

To deploy agentic AI responsibly and effectively in the enterprise, organizations must progress through a three-tier architecture, Foundation tier, Workflow tier, and Autonomous tier where trust, governance, and transparency precede autonomy.

From a ML Manager's perspective

A practical handbook for engineers building, optimizing, scaling and operating LLM inference systems in production.

What are the principles we can use to build LLM-powered software that is actually good enough to put in the hands of production customers? - humanlayer/12-factor-agents

Uncertainty Estimation is an essential ML method for improving robustness in ML models, and building trust/confidence in AI users. In the era of LLMs, it is even more critical because LLMs can be very confident on answers that are very wrong!

My team from…

Best AI teams obsess over measurement and iteration.

If you streamline AI evals, everything else becomes easy.

But we can't take CI/CD from traditional software or ML.

Why?

LLMs introduce unique challenges:

- Non-determinism, no single "right answer"…

Table of Contents Motivation Optimization goal of GPUs Key concepts of GPUs - software and...

If I were building a tech company's internal AI systems from scratch, this is what the architecture would look like:

1. LiteLLM or OpenRouter
This is the bedrock. Allows you to both unify multiple LLM providers behind a single system, while also giving total…

A collection of the most important Github repos for ML, AI & Data science practitioners - GitHub - youssefHosni/Awesome-AI-Data-GitHub-Repos: A collection of the most important Github repo...

This is a dramatized version of how things "may have" unfolded, intended to make core math, engineering concepts, and complicated ideas more interesting to read. All "fictional" content is written in italics so you can differentiate fact from fiction.