I build AI-native platforms and developer tooling myself, end to end
I build production AI systems and developer tools — then I use them myself.
From Job Search Studio's CQRS-driven job-matching pipeline to Compliance Studio's inline AI review agents, I design systems around real workflows, and I ship them end to end.
Job Search Studio didn't feel real to me until it survived a dedicated defect-hardening pass — NestJS, CQRS, Angular, and no shortcuts left in by the end.
I got tired of compliance review living in a separate CI gate, so Compliance Studio puts the same review agents inline in the editor instead.
I built Platform IAM once, properly, after watching three products each start to reinvent OIDC and RBAC slightly differently — now everything I ship here uses it.
Prompt Studio and Chat Panel share one chat-participant contract; the model-routing backends underneath are still separate, and I'm fine with that for now.
Flagship projects
Selected systems that show how I approach architecture, workflow design, and technical execution.

Job Search Studio
A SaaS application that automates job discovery, tracking, and application workflows, built with NestJS, Angular, and CQRS.

Platform IAM
A shared identity and access management microservice — OIDC flows, policy-based RBAC — underpinning every product in this platform.

Prompt Studio
A VS Codium extension for authoring, testing, and iterating on LLM prompts directly inside the editor.
How I work
Architecture
I try to bring DDD, CQRS, and reliability thinking in from day one rather than bolt it on later — Platform IAM only became shared infrastructure because I'd already watched three products start to solve the same auth problem slightly differently.
AI systems
I built the multi-provider AI router that Job Search Studio's ranking and evaluation pipeline runs on, and reused it rather than writing a second one for Prompt Studio.
Developer tooling
I build the VS Codium extensions I actually use myself day to day — Prompt Studio and Compliance Studio both started as friction I was feeling firsthand, not a market gap I'd read about.
Product execution
Job Search Studio only counts as shipped in my book because it went through a dedicated defect-remediation pass first — I'd rather say that took longer than expected than call an unfinished app done.
About

I'm a software architect and technical educator who has spent my career building production systems and writing about the disciplines that separate successful software products from expensive technical failures — lean product discovery, acceptance test-driven development, clean architecture, and data-driven adaptation.
In my day-to-day work, I design and build AI-native platforms and developer tooling: multi-agent systems, prompt engineering workflows, and the infrastructure that lets small teams (or one person) ship production-grade software solo.
This site is where I write that experience down — practitioner notes on the Articles page, and longer-form work in the Books and Courses I'm developing. I'd rather share what I've actually learned building real systems than repeat theory I haven't tested myself.
Current focus
Right now I'm especially interested in AI-native developer tools, local and hybrid LLM workflows, domain-specific interfaces, and systems that help experts work with more clarity and less friction.
- Building AI-native tools for experts working in complex domains
- Exploring local LLM workflows and evaluation layers
- Open to collaborations around DevEx, AI infrastructure, and applied agent design
If you're building something like this, I'd like to hear about it
I'm most interested in conversations around AI infrastructure, developer tooling, domain-specific platforms, and technical products that need both architectural depth and product clarity.
Get in touch
The best ways to reach me:
Or send a message
Have a project, role, or system worth discussing? Tell me a bit about it and I'll get back to you.