Headless SDXL: ControlNet and IP-Adapter Without a WebUI
How I pulled SDXL, ControlNet and IP-Adapter out of the AUTOMATIC1111 WebUI into a reproducible headless CLI pipeline with LoRA, DreamBooth and FID scoring.
Read the noteNotes From Production
Long-form write-ups on the systems I ship: LLM extraction pipelines, retrieval on pgvector, model routing and cost control, headless diffusion tooling, and what forward deployed engineering looks like from the inside. Every external number here carries its source.
LLM Model Routing to Keep Inference Costs Predictable
How I route every AI call in Opaeron through one gateway: task-based model selection, cross-provider fallback chains, JSON repair and per-run cost tracking.
Read the noteMeeting Transcript to Action Items: an LLM Pipeline
How Opaeron turns a pasted client meeting transcript into action items with owners and due dates: chunking, extraction, dedup, and a text guard.
Read the noteSupabase to Self-Hosted Postgres: A Production Migration
How I moved Opaeron off Supabase onto self-hosted PostgreSQL 16 with pgvector: a psycopg3 query-builder shim, losing RLS, pooling and additive migrations.
Read the noteForward Deployed Engineer: What the Job Actually Is
What a forward deployed engineer actually does: embedding with the users, writing production code against the real workflow, and owning it after launch.
Read the note