About
Architecture Through
Operations
I'm Lisa M. Kaminski, a Solutions Architect and AI Integration Specialist. I architect, build, deploy and operate production systems end to end, for clients across North America.
The clearest example is one I own. Dauphin Lake Watch is a live public monitoring platform I designed, built, deployed and have run continuously — through a record snowpack melt and a severe convective event — while a community used it to decide what to do about their property. Building a system and operating one under load are different skills. That one needed both.
Work at this scope usually takes a team. The advantage when it doesn't is that nothing gets lost between the person making the architectural call and the person who has to live with it.
Before this: thirteen years at BellMTS. Budgets, infrastructure investment, and a product I conceived and built that retired paper documentation from more than 3,000 desks — it won organization-wide adoption almost overnight, secured $500K in funding, and grew from one person to thirty-five. I learned what matters to a business by being accountable for it.
I understand the technical side and the business problem, and which of the two is actually the constraint. That's what makes the difference.
Certifications
How I Work
SEE → SOLVE → SCALE
SEE
I make sense of the situation first — the system, the constraints, the missing facts — then map the complete architecture where others see only problems.
SOLVE
I build working prototypes that prove concepts — not strategy documents that gather dust. And choosing what not to build is half the work.
SCALE
From prototype to production-ready systems — failure modes designed first. I turn what's working into what's winning.
What I Bring
Built, Deployed, and Still Running
End-to-End Ownership
Concept, architecture, build, deployment, and the operations afterward — one person, no hand-offs, no gap between whoever designed it and whoever keeps it running. Most work at this scope needs a team and a project manager to hold it together. This doesn't.
Systems Where Failure Has Consequences
A community reading a live monitor to decide whether to sandbag. A regulated practitioner whose licence depends on the records being right. A platform where shipping to the wrong jurisdiction is a violation, not a bug. That is a different discipline from shipping a marketing site, and it is the work I take.
Founder-Operator, Not Just Builder
I run a live public monitoring platform of my own — through a record snowpack melt and a severe convective event, continuously, while people used it to make decisions. Building a system and operating one under load are different skills. I have done both, on the same system, at the same time.
Architecture That Outlives the Brief
One platform I architected was commissioned for a company that then re-engineered its entire regulatory position and went national. It absorbed the change without a rebuild. Anyone can build for the requirements in front of them; building for the ones that arrive later is the actual job.
Production AI, With Boundaries
Stack — the assistant on this site — is a governed deployment, not a chatbot bolted on: a defined knowledge boundary, explicit refusal behaviour, and adversarial testing against extraction. Deploying an LLM publicly and controlling what it will say are two different problems. Most people solve only the first.
AI-Augmented Delivery, Directed
Six months of work landing in six weeks is not the model typing faster — it is strategy fixed before code, design parameters approved before implementation, verification throughout, and full directional control at every step. The judgment is mine; the AI is leverage. That distinction is the whole method.
Business Operations Depth
Thirteen years at BellMTS — budgets, infrastructure investment, and a self-initiated product that retired paper documentation from 3,000+ desks, won organization-wide adoption almost overnight, secured $500K, and grew from one person to thirty-five. I have built inside a large organization as well as outside one.
Canadian Compliance & Accessibility
PIPEDA and Canadian data residency, GST/HST including provincial variance, trust accounting, CASL, and WCAG 2.1 AA accessibility — applied, not academic. Strict compliance, invisible to the person using it.
Full Stack, Including the Parts Nobody Wants
React, Astro, Supabase and PostgreSQL, Python, Docker, WebSocket for real-time, deployed on AWS, Vercel and Netlify with CI/CD. And the unglamorous layer underneath: cloud infrastructure, email authentication, DNS, certificates, and migrations that move a system without breaking mail flow or losing a search ranking. That layer is where projects quietly fail.
You don't have to take any of this on faith. Stack — my AI assistant — will tell you what I've built, what I won't build, and whether I'm a fit for what you need.
Ready to discuss your project?