01 — Blog
Writing on security, systems, open-source projects, and working with generative AI.
Effective Agentic Harness Usage
An LLM is a stochastic sequence completion engine that is very good at sounding human. Working well with a coding harness means holding both of those truths at once, and using them deliberately.
Ethical Usage of Generative AI
Generative AI is intellectual property laundering. Where that is theft and where it is a contribution to the commons depends on the domain, so here is where I land on art, open source, contract work, and local models.
02 — Services
Scoped engagements across the stack — from low-level infrastructure to system architecture. If it touches production, it's in scope.
Software Supply Chain Security
Build pipeline audits, dependency risk, provenance and signing, CI/CD hardening.
Network & Infrastructure Security
Architecture review, threat modeling, segmentation, and hardening for production networks.
Distributed Systems Security
Failure-mode and consensus analysis for large-scale, high-availability systems.
Applied Security R&D
Novel or ambiguous problems — research-to-production work when there's no existing playbook.
Don't see your exact problem listed — ask anyway.
03 — About
Ph.D., Computer Science
Ten years building secure, large-scale infrastructure at Google, Square, and RedSeal — with depth in distributed systems, software supply chain security, and network security.
Comfortable owning a problem end to end: from research question to something running in production. Demilich Digital is that same work, available directly.