Production ML: Avoiding Pitfalls and Keeping It Real

Ever spent weeks, if not months, training a near-perfect ML model, only to see it crumble when it hits production? Welcome to my world in 2023. It’s like watching a beautifully crafted sandcastle get wiped out by a tide of reality. Let’s dive into why that happens and

Hey there, I’m Alex Petrov. As someone who builds agent systems, I’ve waded through the nitty-gritty of getting machine learning models ready for production. It’s like watching a kid graduate from college and start their first job. You might think that getting a model to work in a controlled environment means the hardest

Why I Talk About Kubernetes and Agents

Let me be clear: my love-hate relationship with Kubernetes is a tale as old as time. Early last year, I found myself caught in the classic bramble of agent deployment on Kubernetes. Picture me, 3 AM, piping hot coffee in hand, debugging why an agent wasn’t performing as

A Day in the Life of Debugging
I remember a time when I was up to my neck in logs, trying to figure out why our microservices were acting like teenagers who decided curfews were mere suggestions. Outputs everywhere, but nothing seemed to connect. I was chasing ghosts in different logs across multiple services.
That’s

Debugging Agent Chains in Production: A Practical Guide

You know what keeps me up at night? Agent chains running wild in production. One time, I had an incident that cost us an entire week, hunting down a bug that only appeared in production. Debugging agent chains isn’t just a technical exercise—it’s a battle of wits.

Alright, picture this: I’m trying to get to the bottom of why my AI agent is acting up, and it feels like trying to solve a Rubik’s Cube while wearing oven mitts. If you’ve ever been there, eyes glazed over staring at cryptic logs or endless code, you feel my pain. Honestly, the