Data Scientist

Bison Transport•Winnipeg, MB

4d•Remote

About The Position

The Data Scientist applies advanced mathematical, statistical, and machine learning methods to operational problems drawn from freight and transportation networks—routing, network structure, utilization, demand and cost modeling, and prediction. This is an analytical role, not a data engineering or MLOps role; the focus is on formalizing ambiguous business questions, selecting and defending appropriate methods, and communicating findings and their limitations to operational and executive stakeholders. Graph-structured data is central to how the business is modeled, so comfort reasoning about networks is a distinct advantage.

Requirements

STEM degree required; Master's or PhD in a quantitative field (statistics, applied math, operations research, physics, computer science, or similar) strongly preferred
4+ years of applied analytical or data science experience (mid–senior)
Strong foundation in statistics, probability, and applied mathematics (linear algebra, optimization, or similar)
Experience with predictive modeling and machine learning, and the judgment to know when a simpler method is the right one
Proficiency scripting in Python or R for analysis—clean, correct code, without an expectation of shipping production systems
Fluency with data manipulation (SQL, pandas/dplyr, or equivalent)
Ability to reason about uncertainty and clearly state assumptions and their consequences
Demonstrated ability to explain technical work to non-technical audiences

Nice To Haves

Graph/network analysis (community detection, flow, shortest-path, centrality) or graph-based ML
Time series, causal inference, or experimental design
Optimization or operations research applied to logistics, routing, or scheduling
Domain exposure in trucking, freight, logistics, or supply chain

Responsibilities

Translate ambiguous operational problems into well-posed statistical, mathematical, or graph-theoretic questions
Design and run analyses including statistical modeling, regression, optimization, simulation, and predictive/ML modeling
Build and validate predictive models, and clearly characterize where they generalize and where they do not
Work directly with real data—cleaning, exploring, and interrogating it before modeling
Communicate methods and results to operational and executive stakeholders, including the reasoning behind methodological choices and their limitations
Distinguish correlation from causation and signal from artifact, and state clearly when the data cannot answer the question