Ottawa, Canada | Data Scientist | Ph.D. Earth Science
Production AI for complex, high-stakes data.
I design and ship machine learning systems across enterprise NLP, LLM and RAG workflows, forecasting, remote sensing, and multimodal sensing programs.
- 400M+ records supported in large-scale classification programs
- POC to Prod delivery across prototyping, orchestration, and production rollouts
- NLP to Sensing experience spanning text, forecasting, vision, and multimodal ML
Focus areas
Built for enterprise delivery, research rigor, and field constraints.
My work sits at the intersection of production engineering and model development: problem framing, data pipelines, evaluation, deployment, and iteration.
01
Enterprise NLP Systems
Hierarchical classification, sentence-transformer fine-tuning, active learning, and scalable inference for messy real-world text.
02
LLM and RAG Workflows
Grounded generation, retrieval design, evaluation frameworks, and citation-aware reporting for decision-support use cases.
03
Forecasting and Optimization
Multivariate demand forecasting, pricing signals, anomaly detection, and repeatable Databricks + MLflow pipelines.
04
Computer Vision and Sensing
Remote sensing, object detection, and multimodal fusion for UAV and satellite workflows that have to work outside the lab.
Selected impact
Recent programs, described at the problem level.
Client-facing engagements are intentionally generalized here so the site shows the work without exposing unnecessary delivery details.
Retail modernization
Large-scale taxonomy automation
Built hierarchical NLP classification and active-learning workflows that supported automation across hundreds of millions of product records.
LLM systems
Grounded briefing generation
Designed retrieval and evaluation pipelines for structured, citation-aware report generation from free-form source material.
Aviation analytics
Forecasting and pricing decision support
Combined utilization, maintenance, supplier, and failure signals into demand forecasting and elasticity modeling workflows.
Multimodal sensing
Field-ready low-contrast detection
Implemented EO, IR, GPR, and magnetometer fusion pipelines for operational detection tasks where single-sensor approaches were not enough.
Selected work
Applied AI case studies and research tooling.
Project 1
Enterprise NLP Classification Platform
Hierarchical NLP and active-learning workflows for large-scale product taxonomy modernization.
Project 2
Grounded RAG Briefing Workflows
Retrieval and evaluation pipelines for grounded report generation and briefing support.
Project 3
Forecasting and Pricing Platform
Databricks and MLflow pipelines for demand forecasting, anomaly signals, and pricing decision support.
Project 4
Multimodal Detection Pipeline
EO, IR, GPR, and magnetometer fusion for low-contrast target detection and real-time inference.
Project 5
Satellite Remote Sensing Pipeline
Sentinel-2 ingestion, best-scene selection, and scalable object and change-detection workflows.
Project 6
QuakeLabeler
Seismic annotation and dataset creation toolbox for AI-ready earthquake research.
Experience snapshot
Built from research depth, delivered in production settings.
2023 - Present
Data Scientist
Build and deploy applied AI systems across enterprise, aviation, defence-adjacent, and sensing programs.
Selected engagements
Contributor and Reviewer
Evaluated frontier AI models in mathematics and physics using research-grade prompts, rubrics, and failure analysis.
2019 - 2023
Ph.D. Researcher, Earth Science
Built open-source ML tooling for earthquake detection, denoising, phase picking, and seismic dataset creation.
Publications
Selected research outputs.
Developing Deep Learning Tools in Earthquake Detection and Phase Picking
Doctoral dissertation on deep learning systems for earthquake detection and phase picking.
QuakeLabeler: A Fast Seismic Data Set Creation and Annotation Toolbox for AI Applications
Tooling for rapid seismic dataset creation and annotation.
Blockly Earthquake Transformer: A Deep Learning Platform for Custom Phase Picking
Interactive training platform for configurable seismic phase picking.
Joint Dictionary Learning and Super-Resolution Network on Seismic Noise Removal
Research on seismic denoising and super-resolution methods.
Contact
Open to applied ML, platform, and research-heavy AI work.
If the problem involves ambiguous data, operational constraints, or model delivery beyond a notebook, that is usually where I am most useful.