I'm creating this page to share with everyone my personal projects around technology.
I've been doing this for a long time, and I work in the industry. Please follow along and enjoy.
The Value of an AI Persistent Memory System
A persistent memory layer is the cheapest infrastructure decision in any AI stack. It pays back in tokens, accuracy, and continuity, and it compounds the longer it runs.
7 min readaiagentsmemorytokensarchitectureShipping a Next.js App to the iOS App Store with Capacitor
How to take a Next.js web application, wrap it with Capacitor, and ship it to the iOS App Store — with full access to the device APIs that distinguish a real app from a glorified browser tab.
7 min readnextjscapacitoriosmobileapp-storeRunning Gemma 4 31B on Cloud Run: A Performance and Cost Report
What it actually feels like to use Gemma 4 31B on Cloud Run GPUs after the deploy is done. Above Grok, below Opus, fast enough for real work, and cheap enough that the cost line stops mattering.
8 min readllmgemmagcpcloud-rungpucostAn after-action on an LLM-driven Kalshi trading bot
It had a 59% win rate. It still lost money. Three lessons about edge, prompt-spec leakage, and why a strong reasoning model is not a trading edge.
6 min readllmkalshitradingpost-mortemA git-backed auto-memory system for Claude Code
How to keep Claude Code's per-project memory store from dying with your laptop — a private GitHub repository, a symlink, and one Stop hook.
5 min readclaude-codememorygitdeveloper-toolsDeploying Gemma 4 on Cloud Run GPUs: A Trip Report
What the codelab doesn't tell you. Five gotchas from an afternoon of getting Gemma 4 31B running on Cloud Run's new RTX PRO 6000 instances.
8 min readgcpcloud-rungpullmvllm