gocrazy
Systems explained until they click.
Sharp engineering deep-dives — runtimes, latency, and the machinery under your services.
Featured
Why gocrazy: cut through folklore until the system’s real behavior is obvious — then write the next note.
- Goroutine Leak Profiles Go 1.27 includes new goroutine leak profiles.
- How LLMs Can Find a Needle in a Haystack In this article, we are going to look at how LLMs can find a needle in a haystack.
- Size-Specialized Memory Allocation Go 1.27 improves performance of small allocations using size-specialized allocation functions.
- Do LLMs Have the Memory of a Goldfish? In this article, we will learn how LLMs handle memory so that they are useful to end users in performing complex tasks that require conversation and holding context.
- LAST CALL FOR ENROLLMENT: Build with Claude Code We’re relaunching Build with Claude Code, a 2-day intensive cohort-based course taught by John Kim, who has trained hundreds of engineers at Meta to use Claude Code in real production workflows.
- LLMs as a Judge: How to Know if Your LLM is Healthy In this article, we are going to look at the process of LLM evaluation in detail.
- EP225: Why Does Git Revert Cause Conflicts? git revert looks straightforward until it throws a conflict
- Learn Claude Code, evals, AI systems, and more: ByteByteGo Live is here Most online courses never get finished (~4% completion). Live cohorts get ~40%, roughly 10x higher. Live courses are the only courses people actually finish. So we’re launching ByteByteGo Live.
- A Guide to Application Networking Basics In this article, we will look at the various aspects of networking in detail.
- How Smart Model Routing Can Cut LLM Costs by 10X Cost reduction isn’t a given. It also depends on the types of requests the application receives, the price difference between models, and how well the routing system performs. In this article, we are going to look at v...
- Built for Reliability: How American Express Processes Payments at Scale In this article, we will try to understand how the transaction runs through such a cell-based architecture and how the payments are processed even when some services are failing.
- How to Deal With Errors and Failures in LLM-Powered Applications Apart from normal processing, the application also sends data to a large language model (LLM). It then uses the model’s response to carry out a task.
- EP224: MCP vs RAG vs AI Agents An AI agent is kind of an AI system where the agent performs the task autonomously and takes the decisions.
- How Databases Keep Their Sanity with Concurrency Control So how do we handle such bugs? This is what we are going to try to answer in this article.
- Your Agent Speaks MCP. Give It a Computer. Sprites are disposable cloud computers. They appear instantly, always include durable filesystems, and cost practically nothing when idle. They’re the best and safest place on the Internet to run agents and we want you...
- Generic Methods Go 1.27 adds generic methods—a highly desired language feature.
- Why Your RAG System Is Only as Good as Its Translator Model In this article, we’re going to look at how this embedding model works in an RAG setup and what makes it such a critical part of the system.
- How to Shrink a Language Model Without Making it Too Dumb Models have grown roughly 100-fold in a few years, while consumer graphics memory has roughly doubled. It’s not just a matter of tightening things up to make them fit.
- What Happens Inside an AI Chatbot Between Enter and the First Word? In this article, we are going to look at this entire journey in detail.
- Background Work: From Cron Jobs to Distributed Systems In this article, we will look at various such strategies to perform background work in detail.
- Go 1.27 is released Go 1.27 adds generic methods, encoding/json/v2 package, uuid package, faster memory allocation, goroutine leak profiles, and more.
- How to Make LLMs 3X Faster In this article, we will look at how speculative decoding works.
- How to Steal an AI Model’s Private Thoughts In August 2026, a team at MATS Research, the ELLIS Institute Tübingen, and the Max Planck Institute for Intelligent Systems wanted to test whether the encrypted reasoning blocks that Anthropic, OpenAI, and Google hand b...
- Why Code Verification Matters More Than Ever in the Age of AI In this article, we will look at how code verification works, why the rise of AI-generated code puts more pressure on it, along with the extremely useful insights from Andrea on what the future may look like.
- EP223: Ollama vs vLLM vs SGLang To use open-weight models on your machine, you have three main options: Ollama, vLLM, and SGLang. But each engine handles requests differently.
- Schema Evolution: Changing the Contract Without Breaking What Runs In this article, we will look at schema evolution and strategies for the same.
- GraphRAG: How AI Answers Questions Hidden Across Many Documents GraphRAG was designed to handle the second kind of questions, and we are going to learn more about it in this article.
- Introducing the pkg.go.dev API Introducing the new programmatic API for pkg.go.dev, allowing developers to fetch package and module data directly.
- The New American AI Model Designed to be Customized In this article, we will work through the various choices Thinking Machines made while building Inkling.
- Waymo vs Tesla: Two Ways to Build Self-Driving Cars In this article, we will take a look at both approaches.
- EP222: What is Google’s TPU? A TPU (Tensor Processing Unit) is Google’s custom AI chip, designed from scratch for the giant matrix multiplications that modern models live on. GPUs were built for graphics first.
- A Detailed Guide to API Composition Techniques In this article, we are going to dive deep into the area of the API composition problem and the patterns associated with it.
- GitHub vs Vercel vs Replit: What Dev Platforms Do When AI Code Is Cheap AI models have solved the writing code part of software development to a great extent.
- How Cloudflare Is Making AI Pay for Content In this article, we will go through Cloudflare’s solution in the following five steps.
- How to Fight Clickbait: Meta, LinkedIn & YouTube Case Studies What does it take for a social media platform to stop rewarding clickbait content?
- The Read Path versus the Write Path: Strategies and Techniques In this article, we will look at read path and write path operations and techniques in detail.
- //go:fix inline is an API migration tool Go’s source-level inliner is not about speed — it is a safe way to rewrite call sites when you deprecate an API.
- A cloud that is also an agent workplace Platforms that host apps and agent sandboxes share one constraint — isolation must be a product feature, not a blog promise.
- Agent brains vs hands — separate the loop from the shell Keep the agent process durable; run risky commands in a disposable sandbox you can checkpoint and burn.
- Disposable sandboxes speaking MCP MCP is just a tool protocol — the interesting part is where those tools are allowed to run.
- How Big Models Teach Small Models to Be Smart In this article, we will walk through the idea from the ground up.
- LLM security is a threat surface map, not a prompt tip Treat model-backed products like networked systems — inventory trust boundaries before you ship tools.
- Type construction and cycle detection (Go 1.26) Recursive types are easy to write and hard to validate — Go 1.26 tightens construction and cycle checks.
- Why LLM memory gets expensive (and what to cut first) Context windows are not free RAM — every token you keep multiplies cost, latency, and failure modes.
- pkg.go.dev now has a real API — stop scraping docs The pkgsite HTTP API is GET-only and precision-first — ambiguous module paths fail instead of guessing.
- //go:fix inline and the source-level inliner How Go 1.26's source-level inliner works, and how it can help you with self-service API migrations.
- Building Agents that Don't Break Themselves Building agents is fun. Rebuilding agents that break themselves… less so. A lot of Fly people are building agents with less of a penchant for self-destruction by teaching their agents to do anything risky in a Sprite....
- Hiring: Part Time Instructor, Write Production Grade Code with AI We’re hiring a part-time instructor for “𝐖𝐫𝐢𝐭𝐞 𝐏𝐫𝐨𝐝𝐮𝐜𝐭𝐢𝐨𝐧 𝐆𝐫𝐚𝐝𝐞 𝐂𝐨𝐝𝐞 𝐰𝐢𝐭𝐡 𝐀𝐈”.
- LLM Security Basics: The Full Threat Model In this article, we try to build a map of the full attack surface that threatens an LLM’s security.
- Turn And Face The Strange We’re Fly.io, a public cloud platform that is both our favorite way to put an app on the Internet and our favorite way to safely let a frontier agent coding harness cook. This is a post about our company, the future, a...
- Type Construction and Cycle Detection Go 1.26 simplifies type construction and enhances cycle detection for certain kinds of recursive types.
- Unfortunately, Sprites Now Speak MCP Sprites are disposable cloud computers. They appear instantly, always include durable filesystems, and cost practically nothing when idle. They’re the best and safest place on the Internet to run agents and we want you...
- Why An LLM’s Memory Gets Expensive and How to Fix It In this article, we will learn how LLMs use memory, how it gets expensive, and how to fix it.
- Cache stampedes without the folklore Why a cold key can melt a backend — and the boring patterns that keep thundering herds from forming.
- Escape analysis without the folklore What the compiler actually decides when a value lives on the stack vs the heap — and how to read the evidence.
- How schedulers wake sleeping work A practical look at runqueues, parking, and why latency spikes hide in "idle" systems.
- Token budgets with a four-character rule of thumb Why packing context often divides length by four — what that approximation buys you, and where it quietly lies.
- Ranking agent memories without a neural scorer How a small composite score — kind weights, stored relevance, keyword overlap, capped access — picks what fits a task budget.
- Mining file paths from free-form text How a small regex loop turns prose into path-like strings — with dedupe, length gates, and a hard cap of ten.