Building IncidentOps AI: What I’m Building and Why
A living overview of an evidence-driven incident triage and diagnosis system, the engineering principles behind it, and what I learn while building it.
Writing
Engineering systems,
explained from first principles.
Long-form notes on performance, architecture, intelligent mobile systems and the engineering tradeoffs behind production software.
Start here
Latest writing
Building IncidentOps AI: Measure Retrieval Before You Tune It
What a near-whole-document chunking winner taught me about evidence labels, retrieval metrics, and the cost hidden behind a high score.
Building IncidentOps AI: The Backend Decisions That Came First
Learning Python backend engineering through API contracts, concurrency, persistence, and deployment.
Building IncidentOps AI: Valid JSON Isn’t Enough
Learning to build structured LLM interactions with explicit uncertainty, tested failures, and evidence about their limits.
How Search Works in Production: From Inverted Indexes to Tail Latency
A systems-level tour of lexical and semantic retrieval, multi-stage ranking, distributed search, freshness, and the latency tradeoffs that shape production systems.
Production-Grade AI Applications on Android: An End-to-End Architecture
A technical deep-dive for experienced Android engineers building their first serious AI-powered product.
Prefer a feed reader? Subscribe via RSS.