Data Scientist & ML Engineer · Toronto, Canada
-
2021 — Present
Independent Data Scientist & ML Engineer
Signal · Toronto & Remote
End-to-end ML engagements for startups and scale-ups: problem framing, modeling, deployment, and monitoring. Shipped 60+ models across risk, personalization, NLP, and forecasting, driving an estimated $8M in incremental value.
-
2018 — 2021
Lead Data Scientist
Verde Pay · Fintech
Built and led a 5-person ML team. Designed the real-time fraud platform (0.94 AUC) that cut losses 38%, and stood up the company's first feature store and drift-monitoring stack.
-
2016 — 2018
Machine Learning Engineer
Atlas Commerce · Marketplace
Owned recommendation and ranking for a marketplace serving 40M monthly users. Lifted conversion 24% with a two-tower recommender and rebuilt the experimentation framework.
-
2014 — 2016
Data Analyst & Research Assistant
University Computational Physics Lab
Built simulation and statistical pipelines for high-energy physics datasets, co-authoring two peer-reviewed papers on anomaly detection in noisy signals.
M.Sc. Computational Physics
University of Toronto — thesis on statistical learning for particle-collision anomaly detection. Graduated with distinction.
B.Sc. Computer Science & Statistics
McGill University — joint major, Dean's Honour List, undergraduate research in probabilistic modeling.
Questions, answered.
Full-time roles, fractional ML leadership, and project-based work — from a focused proof-of-concept to building and deploying a production model end to end. I'm happiest when there's a real metric to move and access to real data.
Both. A model that never leaves a notebook isn't a result. I design for serving, monitoring, and retraining from day one, and I'm comfortable owning the MLOps plumbing as well as the modeling.
A scoping sprint runs one to two weeks. A focused modeling-to-production project is usually six to twelve weeks depending on data readiness. Ongoing partnerships are monthly. Clean, accessible data is the single biggest factor in speed.
Almost certainly. I work across Snowflake, BigQuery, Spark, dbt, and the common cloud ML platforms (AWS, GCP, Azure). I adapt to your tooling rather than forcing a rebuild, and I document everything I touch.
I sign NDAs as standard, work within your environment wherever possible, and follow least-privilege access. For regulated data I'm experienced with anonymization, PII handling, and audit-friendly, reproducible pipelines.
Want the long version?
Download the one-page CV, or get in touch and tell me about the problem you're trying to solve.