Positions

Synod IntelliCare, Toronto, ON
CTO & Founding Engineer, 04/2026–09/2026
Startup building tools that check healthcare AI for bias

• Architected and led development of DDFA, a fairness auditing platform for healthcare AI, taking it from first design to a live cloud product in a few weeks. Built as a FastAPI/Python and React/TypeScript system with Canadian census-based bias benchmarking to test whether AI treats patient groups fairly, and deployed on AWS (EC2, RDS, S3, IAM).
• Made the key technical calls on design, security and build priorities, protecting sensitive health data with OAuth2/PKCE authentication, role-based access control (RBAC) and full audit logging.
• Built and directed a 5-person engineering team across backend, infrastructure and frontend, setting one shared technical spec and establishing structured processes (SRS gap analysis reviews, task tracking, cross-functional syncs) so a fast-moving startup team moved in the same direction.
• Partnered closely with executive leadership on company strategy, product positioning and go-to-market priorities, translating high-level vision into an actionable execution roadmap for the team.
• Used AI assistants across design, coding and review to ship about 30% faster, keeping quality high through peer review and automated checks.
• Secured research and accelerator partnerships with the Vector Institute and Sheridan College, strengthening the company’s credibility with investors and academic groups.

Self-employed, Greater Toronto Area, ON
Consultant & Strategic Advisor, 11/2025–Present
Independent advisor helping startups make better decisions with AI and data

• Built budgeting models that show where to spend and where to cut, reducing planned costs by about 15%.
• Advise an early-stage startup on an AI modeling paradigm for efficient asset allocation and portfolio optimization, building models that recommend how to split money across investments, balancing return against risk, and tested on 10 years of market data.
• Use AI assistants to capture the nuance in a founder’s question and build a working model in days instead of weeks, then pressure test it on real-world scenarios before advising.
• Guide decisions by combining decision math (operations research) with machine learning, so the founders get clear recommendations they can act on with confidence.

University of Ottawa, Ottawa, ON
Researcher, 09/2024–09/2025
Research on how people and AI should share important decisions

• Designed a context-sensitive Human-AI orchestration framework, using a multi-stage methodology to systematically assign optimal collaboration modes (HITL, HOTL, HITLFE, HOOTL) based on decision criticality and latitude. The resulting decision guide shows when AI should act on its own, when a person should review, and when a person should stay in charge, based on how risky the decision is.
• Designed Human-AI responsibility allocation using the 4C model (Communication, Coordination, Cooperation, Collaboration) to distribute tasks across interaction layers, defining who is responsible for what between people and AI, ensuring accountability and reducing overreliance on AI and risks like automation bias in high-stakes environments.
• Gave leaders a practical way to decide where to trust AI and where to keep human judgment.

Self-employed, Greater Toronto Area, ON
Consultant, 10/2023–09/2024
Independent practice building decision models for businesses

• Built math models that recommend the best staffing plans, reducing overtime and idle time for a services client.

Argonne National Laboratory (ANL), Lemont, IL
Postdoctoral Appointee, then Research Consultant, 10/2018–06/2023
R&D software for high-energy physics in a U.S. national research lab

• Developed and implemented mathematical and algorithmic techniques for approximating expensive functions in High Energy Physics (HEP), and for directly fitting Monte Carlo simulations to experimental data or data observed in nature using a stochastic trust-region optimization algorithm.
• Built MÆSTRO (Multi-fidelity Adaptive Ensemble Stochastic Trust Region Optimization), an open-source tool that tunes slow, noisy computer simulations to match real experiment data, checking each step so results can be verified. Also developed, maintained, and published Apprentice and Outer Optimization, HEP analysis packages for efficiently constructing polynomial/rational approximations (fast stand-ins for slow simulations) and for performing χ² minimizations.
• Developed robust optimization and design of experiment formulations to decide weights of importance with the goal of tuning a HEP Monte Carlo simulator, replacing personal judgment in weighing competing goals with a repeatable method so tuning decisions could be explained and defended.
• Packaged the methods into a workflow package that efficiently solves derivative-free stochastic optimization problems in a high performance computing environment (supercomputers), so other scientists could reuse them.

George Mason University (GMU), Fairfax, VA
Graduate Research Assistant, Ph.D. Candidate, 2013–2018
Doctoral research on better decisions in manufacturing

• Designed and developed reusable mathematical models for non-linear, stochastic, hierarchical, and temporal manufacturing processes from real-world data, so engineers could apply them to new problems.
• Designed and developed one-stage stochastic optimization algorithms based on deterministic approximation heuristics.
• Developed and published SODA (Stochastic Optimization based on Deterministic Approximations), an open-source method that efficiently finds the best control settings for stochastic processes in a large manufacturing service network with uncertain outcomes, by swapping the uncertain parts for simple estimates.

National Institute of Standards and Technology (NIST), Gaithersburg, MD
Guest Researcher, 2015–2017
Manufacturing modeling R&D in a U.S. government standards lab

• Designed and populated a repository of reusable mathematical models sourced from real-world data, published research and crowdsourced data.
• Developed a software framework and prototype (FactoryOptima) to perform composition, analysis, and optimization on reusable models, letting engineers compare factory designs and trade-offs.

Los Alamos National Laboratory (LANL), Los Alamos, NM
Research Assistant, then Research Technologist, 07/2008–06/2012
Bioinformatics R&D in a U.S. national research lab

• As Research Technologist (2010–2012), designed and developed scientific algorithms for highly variable and large scale bioinformatics tools, and developed and debugged multiple backend modules of the HIV project, helping scientists study HIV and influenza.
• Redesigned the HIV project website using a Model-View-Controller (MVC) framework and web services, so researchers could find and use data more easily.
• As Research Assistant (2008–2010), designed and developed the Tree Viewer, Pruner and Decorator tools to perform selection and annotation of Influenza sequences, and designed a schema for a large Influenza Sequence Database, which taught me to turn messy data into usable tools.

George Mason University (GMU), Fairfax, VA
Graduate Teaching Assistant, 08/2012–12/2012

• Taught and conducted labs for an algorithmic programming class of undergraduate students.