tskulbru.dev | Development, Cloud Architecture & DevOps Blog

Token budgets, sandboxes, and who gets the frontier model.

How I built PadelIQ, an iOS app that turns phone-against-the-glass padel footage into positioning heatmaps and coaching verdicts — with the whole computer vision pipeline running on-device via Core ML and YOLOv8.

Ten years of homelab entropy: a 2009 dual-Xeon running every VM, three EOL Ubuntus, two EOL Raspbians, and a compose file that lies about what it runs.

Anthropic and OpenAI sell flat-rate coding seats and expose zero per-user token counts to go with them.

tmux-resurrect brings your windows back, but every Claude Code pane comes back as a blank chat.

Concurrent Claude Code sessions eagerly spawn a fresh copy of every stdio MCP server at startup.

This week's tech layoffs aren't just about cost-cutting.

A practical account of migrating ~40 Go microservices from individual GitHub repositories into a single monorepo using Go workspaces, path-filtered CI, and independent release management.

How I built Plask, a multi-tenant SaaS that connects to your Google Analytics properties and uses statistical anomaly detection and AI-generated weekly digests to surface what matters.

How I built Mockingjay, an iOS app for secure video recording with real-time encrypted cloud backup.

Why companies should democratize AI access beyond technical teams, and how getting your entire organization using LLMs creates compounding competitive advantages.

How I built Stao, a cross-platform standing desk reminder app for iOS, Android, macOS, Windows, and Linux.

The story of building a Tauri-based HTTP client that treats .http files as first-class citizens, and what I learned along the way.

How I moved from centralized API clients to colocated HTTP files with Kulala.nvim, and why this workflow fits microservice development better than Postman ever did.

A practical guide to extending Claude Code with custom skills, slash commands, subagents, MCP integrations, plugins, and hooks to match your specific development workflow.

AI coding assistants are changing how we write software, but for junior developers, generating production code with AI may be trading short-term productivity for long-term career stagnation.

A comprehensive guide to implementing production-grade Android release pipelines using GitHub Actions and Fastlane, covering beta distribution through Firebase and production releases to Google Play Store.

A practical guide to implementing multi-environment release management using GitHub Actions shared workflows, repository dispatch events, and GitOps patterns for Kubernetes deployments.

How I lost dozens of minutes a week to copy-pasting Slack conversations into GitHub, and built an AI tool to solve it.

Implement production-grade resilience patterns in Kubernetes microservices using circuit breakers, intelligent retry mechanisms, and chaos engineering to handle failures gracefully and maintain system stability under adverse conditions.

A practical guide to implementing database schema migrations using golang-migrate, with MongoDB examples and production deployment strategies using Kubernetes init containers.

Why I decided to use Bluesky as the comment section for my blog, and what it feels like to let the conversation happen out in the open.

Learn how to implement intelligent failover and failback mechanisms for Azure Spot VMs in production AKS environments, achieving up to 90% cost savings while maintaining high availability through automated orchestration between spot and on-demand resource…

Learn how to leverage Azure Spot VMs in AKS to reduce costs by up to 90% while maintaining production reliability through proper configuration and workload scheduling.

Understanding Android's fitSystemWindows property and how to properly handle system UI components like status bars and navigation bars in your Android applications.