skip to content
the ducklabs
home / /now · current focus & opportunities

What I'm building /now.

selective Principal / Staff AI & platform engineering conversations
● Status feed · 5 threads
Building Della: bounded internal autonomy, durable Work, secure read-only browser acquisition, and autonomous imaging.
Evaluating Real read-only autonomy history before any expansion of standing authority.
Writing Durable agent architecture: authority, verification, recovery, and browser security as separate contracts.
Reading Agent evaluation, browser-agent security, durable task semantics, and model-first generative runtimes.
Selective Principal AI Engineer, Principal Platform Engineer, Staff / Senior Staff AI, and AI Platform Architect roles.

01Building

  • Della autonomy — verified read-only browser acquisition, bounded internal objective adoption under standing authority, durable Work/VERIFY/recovery, private evaluation, and autonomous imaging through WebbDuck.
  • ExoBrain — authenticated household runtime, operator control plane, privacy/deletion workflows, living documents, and remote two-runtime document parity.
  • WebbDuck — SDXL, FLUX, Krea 2, and Qwen behind a model-first runtime with cross-architecture Identity / Persona workflows.
  • JobFinder — evidence-grounded resume/cover-letter artifacts plus deterministic ATS automation, rerender verification, and recovery recipes.

02Evaluating

  • How Della's read-only internal objectives behave in real operation: proposal quality, duplicate pursuit, authority/capacity blocks, criteria quality, terminal evidence, and communication policy.
  • Where deterministic ATS adapters stop paying off and bounded browser-agent recovery becomes the right escalation layer.

03Writing

  • What bounded autonomy looks like when authority, durable work, verification, recovery, and communication are independent contracts.
  • What it takes to turn a single-model creative app into a model-first multi-backend platform without fragmenting the product.

04Reading

  • Agent evaluation, durable task/recovery semantics, browser-agent security, model runtime isolation, and architecture patterns for AI platform teams.
05 · Open to the right fit

Principal / Staff work with real technical ownership.

I’m looking for hard AI and platform problems where architecture, implementation, and production accountability stay connected.

Own Architecture end to end System boundaries, state, deployment, security, failure modes.
Build The critical path Stay hands-on through prototypes, integration, and hard implementation work.
Ship Production consequences Reliability, observability, evaluation, recovery, and adoption.
Roles Principal AI & Platform AI Platform Architect · Staff / Senior Staff where the ownership is real.
Consulting $110+/hr Depends on scope and responsibility · limited concurrent capacity.
Location Remote preferred Woodland, WA / Portland metro · relevant travel is fine.
Résumé Get in touch
"Now" pages are inspired by nownownow.com — a snapshot of current focus, not an archive.