At Strong, I build core AI/ML solutions for companies of all sizes. We have applied data science and machine learning to every industry imaginable, taking our partners from data infancy to industry-leaders in data-driven decision making.In June 2024, Strong Analytics was acquired by OneSix to deliver a seamless blend of data and AI solutions.

I deliver AI consulting services to clients based on industry best-practices and technical expertise. I also guide and supervise junior resources to develop their technical and business skills, and coordinate our team's delivery of these projects.I lead AI development projects with cross-functional teams to ensure client success in creating ML models that generate business value. Together, I also work with Amii's team to build our client's confidence in ML technology through teaching and discussions during problem discovery and whiteboarding sessionsI also research and develop novel methods for industrial control using RL.

I developed and tested algorithms for detecting anomalies in machines, towards the ultimate goal of machine intelligence. This research was done in conjunction with industry sponsors, Prof. Johannes Günther, and Prof. Osmar Zaïane at the University of Alberta. Skills: Python, C/C++, MATLAB, hyperopt, MongoDB, PyOD, LaTeXUsing ideas from reinforcement learning, I developed novel algorithms for detecting anomalous behavior in time series. I compared these methods with existing outlier detection algorithms, using global optimization techniques for finding optimal parameters. Data sources include empirical datasets from machinery in the lab, as well as simulators based on the same machinery. Simulator design was based upon first-principles derivation of the dynamical models, as well as data-driven search for unknown material/mechanical properties. To improve performance, I built C/C++ accelerator modules for existing Python code for up to 30x runtime speedups. I regularly communicated results to academic peers and industry sponsors through presentations and written reports.

I tested and evaluated numerous outlier detection algorithms on systems under lab-simulated failure conditions.Skills: Python, C/C++, scikit-learn, TensorFlow, KerasFurthermore, I studied the feasibility of transferring failure knowledge between machines: where information of a certain type of failure on one machine, can be used to teach a similar machine to detect the same types of failures.

I developed a Pyomo extension to handle robust optimization problems in process engineering.Skills: Python, LaTeXI also edited and typeset lecture materials for an upcoming graduate course in stochastic optimization theory.