Running my life like a small platform
A second brain is a platform with one user: an inbox for ingest, skills as jobs, cron for scheduled runs, guardrails for the AI, and git for rollback. An intro to how mine works, where the metaphor breaks, and why its structure keeps changing.
When a NAT Gateway costs more than your database
NAT Gateway costs aren't a networking bug - they're an ownership failure. How to eliminate the AWS egress tax using free Gateway Endpoints, PrivateLink routing, and explicit ADR rules.
Running a DevOps career on cron jobs (and a part-time Master's)
Balancing a full-time DevOps role with a part-time Master's in AI without burning out: deliberate boundaries, async communication, and protected downtime.
High cardinality is the silent killer of observability budgets
One innocent tag can explode into millions of billable time series. How dropping high-cardinality labels at scrape time can cut observability spend dramatically.
The silent productivity killer: Local dev environments
If onboarding takes a 30-page wiki and three days of debugging, your local setup is the real bottleneck. Containerize everything and treat DevEx as a core infrastructure metric.
Rust for the Python dev: shifting your mental model
Moving from Python to Rust means unlearning runtime habits - the compiler forces explicit error handling and memory safety, buying you predictable, high-throughput infrastructure code.
FinOps is an architectural constraint, not an accounting problem
Treat cloud cost like latency or availability - a strict architectural constraint baked into IaC with lifecycle policies and automated teardowns, not an afterthought for the finance team.
Applying system architecture to human problems
Scaling a mentorship program follows the same rules as scaling cloud infrastructure: cache common answers, load-balance your availability, and architect away single points of failure.
If you can't explain it to the Board, it won't ship
The hardest part of senior engineering isn't the code - it's translating technical trade-offs into cost, risk, and speed for stakeholders. A framework and ADR blueprint for winning buy-in.
You probably don't need Kubernetes (yet)
We're conditioned to architect for Google-scale on day one - but a boring stack of managed containers serves millions of requests. Adopt Kubernetes only when you hit real breaking points.
Building my site with Jekyll
Skipping the complex cloud-native stack and building a blog with Jekyll and GitHub Pages: zero hosting costs, zero databases, and a frictionless Markdown workflow.
Setting the bar low: One post a month in 2026
Ditching the habit of hoarding half-written notes: a resolution to publish one real post every month in 2026.