LDLet's connect
Applied AI Systems Builder · Healthcare Operations

I build AI systems that support healthcare operations.

Multi-site workflows, event streams, and decision layers — designed to ship, observed in production.

See the production systemAI workflow methodology

Vancouver, Canada · Open to applied AI, AI systems, and operational intelligence roles · Currently shipping multi-site healthcare deployment

AI tools

ClaudeOpenAIGeminiCursorGitHub CopilotAntigravity

Stack

Next.jsTypeScriptPostgreSQLMicrosoft ListsQR workflows

4

Hospital sites · live deployment

800+

Assets under one shared state model

10+

Years in healthcare operations

Aug 2025

Wheelchair tracking — in production since

Currently available

Vancouver, Canada · Open to applied AI, AI systems, and operational intelligence roles

Start a conversation

What I do

What I actually build.

Five capability areas — each one linked to a real system you can read end-to-end.

View all projects

Multi-site operational systems

QR workflows, event streams, and shared state across distributed sites — designed to keep four hospitals on one operational truth.

See the system

AI workflow engineering

Calibrated inference, ensemble forecasting, and decision-gated pipelines — agents do the heavy work inside specified, observable gates.

See the system

Operational intelligence layers

Forecasting and explainable feature attribution on top of operational data — built for planners, not notebooks.

See the system

Equipment & asset workflows

Lifecycle state models, scan-driven transitions, and chain-of-custody — operational software clinicians actually use.

See the system

Production engineering discipline

Specs before code, tests before commits, every diff reviewed — the algorithm, not the vibe.

See the system

HOW I WORK WITH AI

An algorithm, not a vibe.
This is the loop I run.

Most data-science workflows stop at a notebook. Mine ships into production because every step has a gate — a spec, a test, a review, an observable system — and AI agents do the heavy work inside those gates.

  1. FRAME — WITH THE OPERATOR. Frame the problem with the operator, not the dataset.
  2. SPEC — DESIGN BEFORE CODE. Write the spec before any code.
  3. PLAN — REVIEWABLE UNITS. Decompose the spec into a reviewable plan.
  4. BUILD — TESTS WITH THE CODE. Build with agents — code and tests produced together, every step reviewable.
  5. VERIFY — MULTI-GATE. Verify against a stack of gates — types, lint, tests, eval/regression, secrets check, human review.
  6. SHIP — RELEASE-TAGGED. Ship with a release tag and a written rollback path.
  7. OBSERVE — LOGS & FEEDBACK. Observe in production — logs, traces, and incident feedback drive the next loop.

If the pass check fails, return to Plan and repeat.

Featured work

View all projects
Anchor case · Enterprise deployment
ProductionCase Study

Multi-Site Wheelchair Tracking System

Production asset-tracking platform deployed across multiple hospital sites — coordinating 800+ wheelchairs and clinical assets with QR-driven workflows, lifecycle visibility, and chain-of-custody tracking.

  • Multi-site deployment with role-based coordination
  • QR-driven workflows for intake, dispatch, and return
  • Lifecycle visibility from acquisition to retirement
  • Real-time chain-of-custody across hospital units

Power Platform · Microsoft Lists · React · TypeScript · QR systems

Read the case study

Deployment flow

Sites
QR Scan
Workflow
Lifecycle DB
Ops Dashboard

4 sites

deployed

800+

assets tracked

Multi-site

coordination

Chain

of custody

FORECAST · BGD72.4y ±1.8
SCHOOLING
INCOME COMP
APPLIED AI / POPULATION HEALTH

Population-Health Intelligence Platform

PrototypeAI-AssistedCase StudyDemo Available

An AI-native forecasting layer over WHO, World Bank, and IMF indicators. Calibrated life-expectancy projections with explainable feature attribution for ministry-of-health planners.

TypeScript, ensemble forecasting, quantile regression, SHAP, isotonic calibration
p(MALIG)0.83
CI 0.77 – 0.88
CONCAVE PTS
PERIMETER
AMBIGUITY FLAG
APPLIED AI / CLINICAL DECISION SUPPORT

Clinical Risk Engine

PrototypeAI-AssistedCase StudyDemo Available

A calibrated inference system over biopsy feature vectors. Returns malignancy probability, CI band, and morphology-level attribution for clinician-in-the-loop triage.

TypeScript, ensemble classifier, isotonic calibration, SHAP, ambiguity-flag triage policy
HEALTHCARE SYSTEMS

EquiTrackr

PrototypeCase StudyDemo Available

Modern healthcare equipment workflow platform for tracking equipment states, operational workflows, maintenance coordination, and hospital logistics systems.

Next.js, TypeScript, Tailwind, PostgreSQL, Prisma
FINTECH & PLANNING

SpendWise

PrototypeCase StudyDemo Available

AI-native budgeting and financial planning platform focused on operational budgeting, transaction tracking, onboarding flows, and intelligent financial planning.

React Native, Expo, Node.js, PostgreSQL, Prisma

Let's architect your next system.

I'm currently shipping a multi-site healthcare deployment and have capacity for one more applied-AI engagement. The best way in is a 20-minute conversation about the system you're trying to build.

Start a conversation

Find me on