Apr 15 · 9 min read · The bug was alive for three weeks. On a normal day it cost nothing. On the day it activated, it nearly took the service down. The pattern was simple. An HTTP handler had to fetch data from three downstream gRPC services and merge the results. The tea...
Join discussionApr 14 · 9 min read · For a while, I thought channels were Go's way of doing message passing. Something like Erlang processes or actors, except with a simpler syntax. That understanding is fine if you're writing tutorials. It is not fine when you've just OOM-killed a pod ...
Join discussionApr 1 · 19 min read · Originally published at adiyogiarts.com Discover how to allocate Kimi K2’s 2 million token window as a data warehouse for customer feedback. Learn token budget strategies, validation frameworks, and when long-context beats RAG. Token Strategy TOKEN ...
Join discussionMar 14 · 6 min read · Introduction Ever wonder what happens when the wall that keeps a single prompt separate from a continuous, 1‑million‑token conversation starts to crumble?With the launch of Opus 4.6 and Sonnet 4.6, that future is now here—each model can hold a whoppi...
Join discussionFeb 13 · 11 min read · I've been thinking a lot about how we talk to AI agents and what happens when the conversation goes long enough. Not in a theoretical sense -- I spent about 10 hours in a single session with Pulumi's Neo agent, and somewhere around hour three, someth...
Join discussion
Feb 12 · 10 min read · Why Traditional Context Propagation Fails in 2025 Legacy approaches to context propagation—custom HTTP headers, thread-local storage, and proprietary correlation ID schemes—break down under modern architectural constraints. Thread-local storage fails...
Join discussionFeb 9 · 5 min read · I recently had a chance to experiment Kiro, Amazon’s Coding Assistant. After Windsurf, Cursor, Roo Code, Claude, Codex and GitHub CoPilot, AI coding assistants are familiar territory to me. They are undeniably useful — but they all tend to improve th...
Join discussionJan 12 · 5 min read · Most developers treat Controllers as simple "Model Binders" (Input → Process → Output). However, to build high-performance or complex middleware-like logic, you must understand the Context—the execution environment surrounding the controller. 1. The ...
Join discussionJan 9 · 11 min read · Concepts Fundamentals and Representation At its core, an LLM is stateless. Every API call begins with a blank slate, and the model retains nothing from previous interactions unless explicitly provided. What we call context is essentially the working ...
Join discussion
