martinuke0's Blog

Diagram of a distributed LLM inference cluster with separate prefill and decode workers linked by a KV transfer bus.

NVIDIA Dynamo: A Practical Guide to the Open-Source Inference Serving Framework

Dynamo disaggregates prefill and decode, distributes KV cache across nodes, and routes traffic intelligently. Here’s what it is, how it works, and when to reach for it in production.

Diagram of a prefix tree being shared across concurrent LLM requests.

SGLang in Production: How Structured Generation Makes LLM Serving Actually Fast

SGLang is a serving stack co-designed around how real LLM programs actually look — structured prompts, tool calls, and multi-turn agents. We dig into RadixAttention, the DSL, and what changes when you deploy it next to vLLM.

Multi-cluster failover diagram showing active and standby EKS regions connected through Linkerd gateways and Crossplane compositions.

Architecting Multi-Cluster Failover with Linkerd and Crossplane on EKS

How to design an active-passive multi-cluster failover pattern on EKS that combines Linkerd’s transparent service mirroring with Crossplane’s composable infrastructure claims.

Diagram of V8's TurboFan pipeline showing bytecode, IR graphs, and machine code stages.

Inside V8's TurboFan Optimizing Pipeline: Tracing JIT Compilation from Bytecode to Machine Code

Trace a JavaScript function from parsed bytecode all the way through TurboFan’s Sea of Nodes IR, speculative optimizations, and final machine code emission, with the concrete phase ordering V8 actually uses.

Abstract diagram of cryptographic key exchange between browser, authenticator, and server.

Implementing WebAuthn Passkeys: Inside the FIDO2 Registration Ceremony

Inside the WebAuthn registration ceremony: how challenges are minted, how authenticators create key pairs, and what your backend needs to verify.