Senior Data Scientist
Rigorous. Our models produce defensible, reproducible, physically plausible numbers that go into regulatory filings and board reports. Every calculation has to be auditable.
Pragmatic. Real industrial process data has gaps, drift, mislabeling, and inconsistency. We solve for the data operators actually have, not the data we wish they had.
Curious. The problems sit across thermodynamics, signal processing, and machine learning. Nobody solves them from one discipline.
This is a senior, hands-on role for someone who can own a customer engagement end to end: take a facility's raw SCADA history, work out how that facility physically operates, and build the detection, causation, and quantification models that hold up against it.
You will work alongside our Lead Data Scientist as a peer rather than under close supervision, and directly with emissions engineers, data engineers, and product leadership. You will also be in front of customers — talking to the operator's own facilities and process engineers about what the data shows.
Facility & Process Understanding
- Read P&IDs and process flow diagrams and work out how a site actually operates
- Reason from a SCADA trace back to the physical process producing it
- Identify emission pathways at a given facility — what vents, what is recovered, what is combusted, and under what conditions
- Apply mass balance, pressure-volume relationships, and gas behaviour to constrain and sanity-check model output
- Build and validate emission event detection, duration estimation, volume quantification, and causation models against real customer data
- Engineer features that encode process and domain knowledge, not just statistical signal
- Design validation strategies where ground truth is incomplete or absent — time-holdouts, physical plausibility checks, engineering corroboration
- Quantify and communicate uncertainty: volumes as ranges with stated confidence, not single numbers
- Handle noisy, drifting, gap-ridden sensor data as a first-class part of the problem
- Own model direction for the accounts you carry, from raw data through to the output a customer sees
- Present findings and methodology to customers' technical teams
- Contribute to the shared modelling approach across accounts — what generalizes, what has to be facility-specific
- Support deployment, monitoring, and retraining of what you build
- Strong oil and gas facility and process understanding. You know how upstream or midstream sites work — tanks and their venting pathways (thief hatches, relief valves, vndapour recovery), flares and combustors, separators, compressors, blowdowns. This is the hard requirement, not a bonus.
- Engineering degree — chemical, mechanical, petroleum, environmental, or process (BSc, MSc, or PhD)
- 3+ years building and shipping models on real industrial, sensor, or process data
- Strong Python (NumPy, Pandas, SciPy, scikit-learn)
- Solid time-series and statistical foundations — anomaly detection, changepoint methods, working with noisy real-world signals
- Experience designing validation where clean labels do not exist
- Works independently. You take an ambiguous problem and a messy dataset and come back with something defensible, without needing the work broken down for you.
- Communicates credibly with both engineers and non-technical stakeholders
- Comfortable in a fast-moving startup
- Deep learning experience, PyTorch preferred
- Probabilistic modelling, uncertainty quantification, or Bayesian methods
- Physics-informed or hybrid modelling — physical constraints inside an ML pipeline
- Direct SCADA or historian experience
- Emissions regulation familiarity — OGMP 2.0, Subpart W, Canadian or US methane regulations
- Cloud data infrastructure (AWS, Supabase/PostgreSQL)
- P.Eng or working toward it
- Data scientists who need the oil and gas domain explained to them — this role supplies domain understanding, it doesn't consume it
- Process or facilities engineers who analyse and hand the modelling to someone else
- Candidates who need clean, labelled datasets to work
- Modellers uninterested in what the data physically represents
- Anyone who needs work broken down into defined tasks before they can start
- Competitive salary and comprehensive benefits
- Genuine ownership — a small team where your work ships to production and reaches major energy companies directly
- Peer-level technical collaboration on physics-informed modelling
- Flexible arrangements — downtown Calgary office or fully remote
Calgary-based, with preference for Calgary-area candidates. Remote options available.
Send a brief introduction explaining your interest in Arolytics, along with your resume, to info@arolytics.com, referencing "Senior Data Scientist."