LLM Arithmetic Reliability Test — 2026-03-23
This test evaluated whether a large language model (LLM) can reliably perform arithmetic by inference — without code execution or a calculator. The results r...
This test evaluated whether a large language model (LLM) can reliably perform arithmetic by inference — without code execution or a calculator. The results r...
This morning, my OpenClaw AI assistant computed my trade journal P&L and got it wrong. Not a little wrong. Obviously wrong. Off by 33%, delivered with co...
AI agents are getting access to credentials faster than your governance processes can track them. Here’s what the NHI explosion actually looks like — and how...
Legion is my AI partner — a self-hosted OpenClaw installation built from source and wired into the tools I actually use.
Building Fake Spock — a Mattermost chatbot tuned to behave like a Vulcan and deliver dad jokes.
From Docker Compose to production Kubernetes with Ansible, MetalLB, and cert-manager.
Self-hosting LiteLLM, Langflow, and Khoj on Kubernetes while using AWS Bedrock for inference.
Migrating AI workloads from AWS Lambda to on-premises Kubernetes to cut costs.
Free TLS certificates with automated renewal using Certbot, Route 53, and GoCD.
You can’t trade context you don’t have — and the fifth pillar of stock selection just landed. MDP now has real-time news, sentiment scoring, and a flame syst...
Running the stock scanner with Docker Compose — setup, configuration, and first launch.
Building a real-time stock scanner with FastAPI, RabbitMQ, and Redis — Part 1.
Building a random dad joke site in an hour with vanilla HTML and zero frameworks.
A two-layer cache-miss prevention strategy using asyncio.Event and Redis locks.
Wrapping up Wave 1 — debugging stories, 1,490 msg/s throughput, and 100% test coverage.
How OpenTelemetry exposed a hidden bottleneck and drove architectural improvements.
A thought experiment on why Brainfuck reveals the fundamental limits of AI-assisted coding.
Why I’m moving my canonical content to a platform I own — and what that means for readers.
Introducing the concept of building impressive tech solutions on a budget.
From the Apple IIe to the USS Nimitz to a 20+ year career in tech — how it all started.
From Raspberry Pi experiments to a full Proxmox cluster — the honest evolution of a home lab.
AI agents are getting access to credentials faster than your governance processes can track them. Here’s what the NHI explosion actually looks like — and how...
AI agents are getting access to credentials faster than your governance processes can track them. Here’s what the NHI explosion actually looks like — and how...