02 / INFRASTRUCTURE
A machine with a job to do.
My current computer handles everyday use and the development work shown on this site. The proposed workstation would give long-running services, AI experiments, and project infrastructure a separate place to run.
Why separate the work?
Development servers, databases, containers, browser jobs, local AI tests, and media processing share the computer I use every day. Keeping them running can interrupt normal use, and normal use can interrupt them.
One personal computer
Everyday use, Dillon HQ, Grant Wizard, automation, browser jobs, servers, and development environments compete for the same resources.

An estimated target for the machine, power protection, and supporting parts. The RTX 3060 is already owned and would be reused. Component prices and substitutions may change the final cost.
Hardware plan
Parallel development and background services
Local model experiments and media workloads
Reuse for separate workloads where supported
Expansion and connectivity
Concurrent services, containers, and databases
Projects, datasets, models, and backups
Electrical headroom
Sustained workloads
Airflow and room for two GPUs
Power protection and controlled shutdown
Complete ready-to-run environment
Not every dollar must be spent. The two GPUs are intended for separate workloads where supported; they would not automatically combine in every application.
A short window to shut down safely.
The UPS is intended for brief flickers and a controlled shutdown, not hours of operation. After a sustained outage, the machine would stop new jobs, save state, close services, and shut down normally. The recovery policy would be configured and tested after the system exists.
How the systems would connect.
This describes the intended relationship after a dedicated machine is in place. The project screenshots elsewhere on this site show current software.
Grant Wizard · Content Factory · Opportunity Research · browser automation · local AI · coding agents · databases · server workloads · background workers · monitoring
A practical path forward.
- CurrentToday
Projects exist on shared hardware
- GoalDedicated compute
Separate development machine
- PlannedReliable infrastructure
Persistent services, workers, testing, and databases
- Aiming forStronger products
More room for development and iteration
- ExploratoryBusiness experiments
Test whether useful projects can become products or services
Send hardware directly.
The Amazon wishlist contains specific workstation components and accessories. Amazon handles delivery, so supporters do not need my home address.