← home / notes

Notes.

On distributed systems, reliability engineering, and what LLM agents are actually good for in infrastructure.

The Service Graph Is a Lower Bound

The dependency graph you draw from RPC traffic is real but incomplete. The dependencies that take you down are…

5 min2026

Streaming Ingest in Rust

When you fuse many noisy event streams into one model, the hard problems are not throughput - they are identity and…

7 min2026

Closing the Loop

A pattern worth reusing: when an LLM makes a judgment at scale, do not let it grade itself. Pair a cheap predictor…

6 min2026

Build the Substrate First

Most platforms that reason about a system are really stacks of queries against a model of that system. The recurring…

6 min2026

Agent Harness vs RAG

Two ways to feed a model context: retrieve fuzzy passages by similarity, or call purpose-built tools that return…

6 min2026

Evals Before Features

Before you wire an LLM into a real workflow, decide how you will know it is good enough - because the eval is the…

5 min2026

LLM as a Judge

A separate model can score outputs you cannot label by hand - but only if you treat it like a measuring instrument: a…

6 min2026

Cost-Aware LLM Pipelines

Most items in a large workload are easy, and a few are genuinely hard. The cheapest reliable pipeline is the one that…

6 min2026

Designing APIs for Agents, Not Humans

When an LLM is the caller, the interface is the prompt. Typed responses, idempotent writes, granular composable…

6 min2026

Your Model Is Not the Product

A correct model or a sharp analysis is necessary but not sufficient. What gets internal AI actually used is the last…

6 min2026

SLO-Driven Risk

Reliability effort gets spent on whatever feels scary in the room. Here is how to replace that gut feel with a…

5 min2026

Scaling a Live Stream to a Billion Viewers

A live broadcast turns one source into millions of simultaneous viewers in seconds. The hard part is not the video -…

4 min2026

Natural Language to SQL, Then and Now

In 2018 I helped build an RNN model that turned English questions into SQL, trained on WikiSQL and later published…

3 min2026