Local Gemma with Hermes: Cheap Delegation for Rails Maintenance
A practical guide to using Hermes Agent with a local Gemma model for routine coding work, while reserving cloud models for tasks that genuinely nee...
Read the article → Joey
Practical notes on software, AI, and the systems around them
I write about building reliable software, learning platforms, and AI-assisted workflows—with the implementation details, trade-offs, and failures left in.
Start here
A short path through the ideas that shape this site.
Golden Rules for Cheaper, Safer LLM Agents
Notes on turning repeated AI workflows into scripts, skills, cron jobs, and guardrails so agents become cheaper, safer, and more useful over time.
Read the article →AI Application Architecture: LLM + Memory + Tools
The five components a production AI application actually needs beyond the model itself: memory, tools, retrieved knowledge, and structured context.
Read the article →Latest notes
Sending Email from an Ubuntu VM on Oracle Cloud (OCI)
A Postfix and OCI Email Delivery setup guide covering SMTP auth, sender approval, SPF/DKIM/DMARC, and the errors that block delivery.
Ruby Ractor vs the GVL: When You Get Real Parallelism
How Ruby's Global VM Lock limits threaded CPU-bound code, and how Ractor bypasses it entirely by trading shared memory for message passing.
Rails in Kubernetes: Migrations, Jobs, and Puma
Running a Ruby on Rails app in Kubernetes means containerizing it correctly and handling migrations, background jobs, and Puma as separate concerns.
Whisper: An Open-Source Alternative to Cloud Speech-to-Text
OpenAI's Whisper is a self-hostable alternative to Google's Speech-to-Text API, with a Flask wrapper and a C++ port covered here alongside the original model.
Ruby's GVL vs Ractor: What Actually Runs in Parallel
Ruby's Global VM Lock serializes threads for CPU-bound work, and Ractor works around it with isolated memory, benchmarked here against threads and processes.
Speeding Up Nested JSON Serialization in Rails
Caching, the Oj gem, eager loading, background jobs, and load_async each cut serialization time for deeply nested Rails associations like classrooms and scores.
Zero-Downtime PostgreSQL Upgrades With Logical Replication
A bash script that sets up PostgreSQL logical replication, publications, and subscriptions to perform a major-version upgrade without downtime.
Redis Memory Spikes: Tracking Down a Runaway Sidekiq Queue
How I traced a Redis memory jump from 130MB to 800MB back to an unthrottled Sidekiq integration, and the maxmemory, eviction, and unique-job fixes that followed.
Scaling PostgreSQL on Kubernetes with Kubegres
How Kubegres brings primary/standby failover, automatic backups, and PITR to a self-hosted PostgreSQL cluster on Kubernetes without a managed database bill.