AI Engineering
Leader.

1. 01

   Made “accurate” defensible

   Facility-disjoint validation, bootstrapped confidence intervals, inter-reader baselines and imbalance-aware model selection became standard practice — not per-project choices.

2. 02

   Trusted nothing unmeasured

   Proved DICOM tags systematically wrong, built ground truth from three-radiologist consensus, and benchmarked a well-marketed foundation model — then kept the in-house one.

3. 03

   Turned one-offs into platforms

   One serving framework, 157 versioned annotation configurations, shared prompt tooling. Each new model cost less to ship than the last.

4. 04

   Built the capability, not just the output

   Hired deliberately from Canadian master's and PhD graduates instead of competing for scarce senior talent — real depth, within a Series A budget.

02/Career Journey

From hands-on ML to building the organization.

1. Jun 2025 — Aug 2026

   Vancouver, BC · US & Canada

   07

   Executive Director, Product Management & AI Development

   Series A medical-imaging company

   • Architected an eight-agent LLM system for radiology report generation, shipped into clinical workflow across five environments — 35% faster report turnaround.
   • Required generative output be measured before release: an evaluation harness with LLM-as-judge scoring, bootstrapped CIs and subgroup breakdowns — near-zero measured hallucination and zero false-positive recommendations on a 330-report stratified benchmark.
   • Led a 15+ person cross-functional org across three time zones at 95%+ on-time milestone delivery; owned 14+ AI products end to end.

   + 2 more− Less

   • Program manager for a $30M strategic enterprise partnership, with formal governance across product, engineering, regulatory and support.
   • Established the Good Machine Learning Practice framework; participated in the FDA 510(k) submission for the XR Chest and XR Spine Finding Detectors.

2. May 2024 — May 2025

   Vancouver, BC

   06

   Director of AI Research & Development

   Series A medical-imaging company

   • Directed AI R&D across X-ray, CT, MRI and clinical NLP, leading 10+ researchers and engineers from ideation to cleared deployment.
   • Cut manual annotation effort 50% with model-assisted labeling — including mining structured supervision from ~31,000 free-text radiology reports.
   • Set the evaluation standard: facility-disjoint validation, confidence intervals, and error benchmarked against measured radiologist inter-reader variability.

   + 1 more− Less

   • Commissioned a build-vs-adopt study of a medical foundation model; the in-house model won, preventing a costly platform commitment.

3. Aug 2022 — Apr 2024

   Vancouver, BC

   05

   Senior Machine Learning Manager

   Series A medical-imaging company

   • Built the AI organization from zero to 14 engineers and scientists — sourcing, assessment, offers and onboarding; new hires shipped production features within two months.
   • Directed consolidation of clinical model serving onto one shared platform: seven model repositories, one CI/CD pipeline, five environments in two regulatory jurisdictions.
   • Made automated verification & validation a release gate — releases blocked unless case matrices passed against live deployments.

   + 1 more− Less

   • Cut training-to-deployment cycle time 45% through automated MLOps (MLflow, DVC) on GCP and AWS.

4. Dec 2021 — Jul 2022

   Vancouver, BC

   04

   Machine Learning Manager

   Series A medical-imaging company

   • Led cross-disciplinary ML teams delivering medical imaging and NLP systems.
   • Evaluated and onboarded three vendor platforms, cutting external tooling costs 20%.
   • Raised team NPS by 25 points across two performance cycles through a culture of rigor and code ownership.

5. Apr 2019 — Nov 2021

   Vancouver, BC

   03

   Senior Machine Learning Engineer

   1QBit

   • Architected and delivered XrAI, a Health Canada-cleared deep learning model for COVID-19 diagnosis from chest X-rays — 60% faster radiologist triage at peak pandemic demand.
   • Authored an ML platform codebase later adapted as the foundation of my next organization's production ML platform.
   • Built GAN- and CNN-based bone age estimation models for pediatric radiology (+18% diagnostic accuracy vs baseline reads).

6. Feb 2014 — Mar 2019

   Doha, Qatar

   02

   Data Scientist

   Sidra Medicine

   • Developed deep learning models for early-stage skin cancer detection — 94% sensitivity in clinical validation.
   • Built ML queuing analytics that cut average wait times 22% across three outpatient departments.
   • Integrated ML into EHR workflows with clinical teams, reducing manual data entry 35%.

7. Sep 2017 — Jun 2019

   Doha, Qatar

   Part-time · concurrent

   01

   Teaching Assistant, Applied Deep Learning

   Hamad Bin Khalifa University

   • Lab sessions and one-on-one mentorship in deep learning, CNNs and TensorFlow for 40+ graduate students.