Brett Wiens

Professional

People Analytics Leader • Data Platform, Enablement, Trust

Hi, I’m Brett.

I build trusted data products that turn ambiguity into action. Sometimes that means transforming legacy spreadsheets into a modern, governed platform. Sometimes it means automating repetitive work so teams can focus on judgement and outcomes. And sometimes it means deploying machine learning and AI in ways that are explainable and actually get used. I’m at my best when I’m making people’s lives easier and helping leaders make confident decisions.

Mission

My goal is to maximise the value data can deliver by combining leadership, modern architecture, and hands-on delivery. I build capability in others, modernise platforms so insight is reliable, and ship solutions that are explainable, trusted, and maintainable.

Read more

I build and lead people first. I mentor and teach across the full experience spectrum, helping individuals use modern tools with confidence so they can grow their own careers and contribute more meaningfully to the organisation.

I guide organisations away from fragile, outdated tools and toward modern platforms and frameworks that support faster, more reliable delivery of data, information, and insight. Architecture matters because it sets the ceiling for what teams can achieve.

I also stay hands-on. I work directly with an evolving technology stack to build bespoke solutions for real business problems, with a strong bias toward designs that can be explained, trusted, and maintained over time. Clear communication and thoughtful presentation are essential to good data work.

I am at my best when I save someone significant time, make their work easier and more valuable, or help them develop the confidence and skills to build solutions they once thought were out of reach.

Journey

Skimmable career arc, with outcomes and credibility signals.

Manager, HR Reporting, Analytics, and Data · Cenovus Energy
2023 to Present
Built a modern, trusted HR data practice that turns manual processes into governed, automated delivery.
HR Data Strategy
Crafted HR data strategy, governance, and platform modernisation, moving the organisation away from fragile spreadsheets and manual processes toward a trusted, governed data platform.
People Leadership
Managed and mentored a team of data specialists, balancing delivery pressure with skill growth, engagement, and long-term capability building.
Workforce Planning
Business Case
Effective workforce planning requires a clear and reliable view of both current and future staffing. The existing workforce planning solution had been delivered by a vendor at significant cost but had become complex, fragile, and time consuming. Planners were spending weeks each month maintaining the system, limiting their ability to focus on higher value strategic planning. The workflow relied on disconnected tools, manual processes, and limited foundational design, creating a large recurring operational burden.
<div style="padding:14px;border-radius:16px; background:rgba(170,160,10
Comp & LRP
Business Case
Accurate compensation forecasting is a critical input to long range and budget planning. By combining workforce plans with detailed compensation data across wages, bonuses, benefits, pensions, and allowances, the organization can forecast future G&A costs with confidence. Historically, producing these forecasts was complex, time intensive, and often required rapid turnaround, placing pressure on both timelines and data quality.
Approach
Working closely with senior financial analysts, the full end to end compensation planning workflow was mapped and redesigned to replace manual steps with automated cloud based solutions. Data was sourced directly from Workday and SAP, and a position level cost model was built to account for employees versus vacancies, scheduled compensation changes, bonuses, benefits, pensions, allowances, and currency conversion across assets. The model was designed to support the full complexity of compensation structures within a global energy company.
Tools / Stack
Python, Databricks, Delta Lake, API integrations, Excel based reporting.
Impact / Value
Converted a difficult, time intensive process into a repeatable, on demand capability. Compensation forecasts can now be produced for any point in time since data collection began and completed within approximately one hour prior to validation, significantly improving responsiveness and planning confidence.
Applications
Designed and built CRUD-style applications to support a variety of HR functions, moving beyond static reporting into transactional, user-driven tools.
People Metrics
Business Case
Providing timely and accurate workforce information to leaders is a core function of HR. Leaders consistently seek deeper insight into their operations to support better decision making. The objective was to deliver reliable, scalable people metrics sourced from Workday that provided meaningful insight while respecting privacy and governance requirements.
Approach
The process began with Excel based Workday reports and manual calculations, which were time intensive and difficult to scale. We transitioned to direct API integration with Workday to automate data collection and implemented all metric calculations in Databricks to eliminate manual processing. As demand grew, automated pipelines were built to support dozens of metrics across hiring, attrition, demographics, workforce planning, and absence, with results delivered through a PowerBI front end in near real time.
Tools / Stack
Workday API, Python, Databricks, PowerBI.
Impact / Value
Expanded from producing a small number of manual metrics for a single team to delivering over one hundred standardized metrics across dozens of teams. Leaders now have near real time access to workforce insights, enabling better informed decisions while maintaining strong privacy and data governance controls.
Sr. Advisor, Data Science · Cenovus Energy
2017 to 2023
Shipped predictive models and AI systems that stood up to scrutiny and created measurable value.
Electricity Demand
Business Case
Oil sands facilities consume significant electricity, with two sites offsetting costs through cogeneration. Alberta’s coincident peak pricing model imposes substantial tariffs based on electricity usage during the highest demand period each month. If cogeneration units are offline during this window, facilities can incur millions of dollars in additional costs. The objective was to predict, as early as possible, when coincident peaks were most likely to occur and provide actionable guidance for maintenance planning.
Approach
Time of day and weather were identified as the primary drivers of electrical demand. Historical demand and multi year weather data from major Alberta municipalities were analyzed, confirming time and temperature as the strongest predictors. Seven day weather forecasts were combined with historical averages beyond that window. Thousands of neural networks were trained using TensorFlow and PyTorch, with MLflow used to track performance. A probabilistic approach estimated the likelihood of coincident peaks by hour and day, with results delivered through a PowerBI dashboard for operational use.
Tools / Stack
R and RStudio, AWS (CloudWatch, SageMaker, S3, RDS, Lambda, EC2), Databricks, Python, TensorFlow, PyTorch, MLflow, PowerBI.
Impact / Value
The model has provided sustained value over multiple years. During one maintenance period, falling temperatures triggered a high probability alert, prompting expedited cogeneration restart and avoiding several million dollars in additional costs that month alone. While avoided costs are difficult to quantify precisely, the solution continues to save millions annually while requiring minimal ongoing maintenance.
Emissions Model
Business Case
Facilities are required to monitor and report nitrogen oxide emissions to remain within approved limits. Compliance is traditionally achieved using Continuous Emissions Monitoring Systems (CEMS), which are expensive to install, maintain, and replace. The opportunity was to develop a predictive emissions monitoring system (PEMS) as an alternative. No PEMS had previously been approved in Western Canada, but success offered the potential to avoid several million dollars annually in capital and operating costs.
Approach
Alternative process sensors were identified and evaluated in collaboration with engineering to develop a physics based understanding of emissions behavior. Over 200 candidate tags were assessed, with approximately two dozen selected as informative. Modeling progressed from linear regression and random forest approaches to neural networks and XGBoost. Given regulatory emphasis on transparency and explainability, a neural network approach was selected. We worked closely with the regulator throughout, iterating hundreds of models, exceeding required performance thresholds, and contributing to updates of the regulatory framework to enable PEMS approval.
Tools / Stack
R and RStudio, Databricks, Python, TensorFlow, MLflow, direct integration with operational technology systems, and custom data visualization and monitoring tools.
Impact / Value
Delivered the first regulator approved PEMS model in Western Canada, demonstrating the feasibility of replacing CEMS with data driven alternatives and unlocking potential savings in the millions annually. During ongoing monitoring, model drift risk was identified. Although still within regulatory bounds, the decision was made to suspend deployment in favor of environmental protection, reflecting strong governance and risk management while establishing a foundation for future PEMS development.
Production AI
Built and deployed machine learning models that shipped into production environments and delivered measurable business outcomes.
Training
Trained staff across the company in R and Python programming, improving data literacy and reducing dependency on a small group of specialists.
Mentorship
Mentored students and new graduates in data science tools, techniques, and professional practice.
Sr. Advisor, Geospatial Services · Cenovus Energy
2009 to 2017
Created cartographic and spatial analytics tools to improve decisions and support numerous teams.
Geospatial Leadership
Managed a team of geospatial analysts supporting corporate groups including mineral land, environment and regulatory, ground deformation, acquisitions and divestitures, investor relations, and communications. Over the course of my tenure in this team, I worked with and managed senior, intermediate, and new grad GIS analysts, supporting any corporate function. I helped lead the geospatial services team along with two peers for several years.
Heritage Royalty
Business Case
Following Cenovus’ acquisition of ConocoPhillips’ Canadian assets, the company carried higher debt than desired. To accelerate debt reduction and refocus capital on core assets, Cenovus chose to divest its fee title royalty lands. Success depended on rapidly producing high confidence, market ready information for buyer due diligence.
Approach
Selected to lead the data component of a time sensitive, highly confidential divestiture due to prior experience with mineral land data and analytical tooling built for the mineral land team. Built a single authoritative dataset by aggregating and reconciling inputs across mineral land, drilling, exploration, accounting, and production accounting, forming the backbone of the data room under aggressive timelines.
Tools / Stack
Microsoft Excel, including extensive macro driven automation. Constrained by the tools available at the time, the solution was engineered for speed, accuracy, and repeatability.
Impact / Value
Enabled accurate, defensible data delivery for buyer due diligence and supported timely execution. The divestiture closed as the largest Canadian E&P royalty transaction in history, generating approximately $3.3B in proceeds and materially contributing to Cenovus’ debt reduction strategy.
Lease Exposure (MLE)
Business Case
Oil sands leases issued by the Alberta Crown include minimum evaluation obligations during the primary term (for example seismic coverage and qualifying stratigraphic or cored wells). Any portions that do not meet the Minimum Level of Evaluation (MLE) can be forfeited back to the Crown at term expiry. This creates significant asset and value risk, making accurate, proactive tracking essential.
Approach
On joining the mineral land team, I found a largely manual workflow: measuring seismic on maps and checking wells one by one, taking roughly nine months each year and remaining highly reactive. I produced geospatial analyses that classified every Alberta section by whether it met seismic thresholds and mapped qualifying wells by type, reducing annual effort from months to about a week. As requirements evolved, we expanded into an automated, spatially aware evaluation program using Python and vendor supplied GeoLogic data, enabling lease evaluation at provincial scale, including competitor tenure, in minutes.
Tools / Stack
ESRI ArcGIS, Excel (pivots and macro automation), Python, vendor data from GeoLogic.
Impact / Value
Reduced evaluation cycle time from roughly nine months annually to about one week, then to minutes at scale. Enabled earlier identification of material exposure, surfacing hundreds of millions of dollars of potential liability and informing proactive planning and budgeting. Improved decision making by contributing feedback into tenure writing and expanding capability to evaluate competitor positions, supporting farm in and opportunity identification.
Patented Innovation
Business Case
GeoPlanNet was introduced as an optimization concept based on advanced network mathematics that showed significant theoretical savings for oil sands facility layout. However, the model relied on unrealistic assumptions, including full reservoir knowledge upfront, fixed future wellpad locations, and uniform surface development costs. In reality, facilities are built incrementally as reservoir knowledge evolves, with infrastructure expansion constrained by both capital and environmental considerations. Bridging this gap presented an opportunity to reduce surface disturbance while optimizing investment.
Approach
We adapted idealized Steiner tree concepts by integrating them with GIS based probabilistic planning. Reservoir uncertainty was incorporated through weighted likelihoods of future wellpad development, allowing higher confidence areas to influence network growth more strongly. Network expansion was guided using weighted centroids rather than direct extensions. Surface routing incorporated cost weighted raster layers informed by environmental and regulatory inputs, encouraging avoidance of sensitive areas while balancing incremental cost. The result was a probabilistic, spatially aware network capable of generating dynamic plans at any stage of development.
Tools / Stack
ESRI ArcGIS, GIS raster analysis, probabilistic spatial modeling, and extensive collaboration with reservoir, development, environmental, and regulatory stakeholders.
Impact / Value
GeoPlanNet provided planners with unique, actionable intelligence for infrastructure development, supporting more environmentally conscious and capital efficient decision making. The methodology was contributed to COSIA due to its environmental protection value, a patent was filed under the GeoPlanNet name, and at least one COSIA partner adopted the approach. Internally, the outputs informed ongoing planning and development decisions across the asset.
A&D Support
Provided spatial and non-spatial analytics support for multiple high-value, confidential acquisition and divestiture initiatives, contributing to valuation, risk assessment, and transaction readiness. I was involved as the spatial lead for all of our A&D activities as well as mineral land sales. I would create maps and review nearby competitors with our competitor intelligence group to predict bids and to optimize our own.
Data & Vendors
Maintained enterprise spatial datasets and partnered with external vendors to design and deliver custom geospatial data products supporting a wide range of GIS and analytics initiatives. Several of the projects mentioned, particularly MLE and GeoPlanNet, required custom data that wasn’t readily available. Our vendors had the necessary data, but it needed to be delivered in a different structure to support our work. I worked very closely with them to build datasets that met our needs, and generate a useful data product for them as well.
GIS Consultant · Golder Associates
2004 to 2009
Delivered GIS analysis, mapping, and automation to support environmental, regulatory, and planning work across diverse projects.
Archaeological Potential Mapping
Developed an archaeological model for Alberta, and in particular the Oil Sands, that included raster calculation based on features like ecological classification, distance from water, slope and aspect, to produce a predictive archaeological model to help our archaeologists in identifying sites of potential interest. The model was presented unofficially and unexpectedly at the ESRI User Conference in a special interest group session on archaeological modeling.
Cartographic Automation
Recognized that one of the most frustrating and error-inducing parts of standard figure editing (map edits), was changing the values in the legend, and in particular the file path information on the maps. I I took it upon myself to learn how to program in visual basic, upon which ArcObjects operated, and I built a custom tool that provided users with a simple interface with which they could update the author, reviewer, and approver names and dates, the map title, and the file path (automatically), to save time, and avoid frustration. I built a little extension called CygniCarta that was used for some time after I moved on, before ESRI included most of those functions directly in the base software.
GIS
The basic building blocks of geospatial analysis for a wide variety of clients. Included: Spatial data preparation, analysis, and mapping in ArcGIS, with a focus on clean inputs, defensible methods, and clear visuals that supported real project constraints.

Impact Portfolio

  • HR data platform modernisation
  • Workforce planning transformation
  • Compensation and people cost modelling
  • Applied AI systems in production
  • Enablement and mentorship across teams

Education

Bachelor of Arts (Honours) · University of Calgary
2004
Honours Thesis: Predicting infill and gentrification in Calgary using demographic and spatial features.
Master of Geographic Information Systems · University of Calgary
2005
Thesis: Analyzing the 2000 Canadian federal election using sociodemographic variables to predict election results.

Certifications & Continuing Education

Deep Learning & Neural Networks
  • Build Better Generative Adversarial Networks (GANs) · Coursera (2022)
  • Build Basic Generative Adversarial Networks (GANs) · DeepLearning.AI (2022)
  • Improving Deep Neural Networks: Hyperparameter Tuning, Regularization, and Optimization · DeepLearning.AI (2020)
  • Neural Networks and Deep Learning · DeepLearning.AI (2020)
  • Structuring Machine Learning Projects · DeepLearning.AI (2020)
Data Science & Applied Analytics
  • IBM Data Science Specialization · Coursera (2019)
  • Applied Data Science Capstone · Coursera (2019)
  • Data Science Methodology · Coursera (2019)
  • What is Data Science? · Coursera (2019)
Programming, Databases & Tooling
  • Python for Data Science and AI · Coursera (2019)
  • Machine Learning with Python · Coursera (2019)
  • Data Analysis with Python · Coursera (2019)
  • Data Visualization with Python · Coursera (2019)
  • Databases and SQL for Data Science · Coursera (2019)
  • Open Source Tools for Data Science · Coursera (2019)
  • Complete Python Bootcamp · Udemy (2019)
  • Ultimate MySQL Bootcamp · Udemy (2019)
Statistical Foundations
  • Statistical Inference · Coursera (2017)
  • Regression Models · Coursera (2017)
  • Exploratory Data Analysis · Coursera (2017)
  • Getting and Cleaning Data · Coursera (2017)
  • Reproducible Research · Coursera (2017)
  • R Programming · Coursera (2017)
  • The Data Scientist’s Toolbox · Coursera (2017)
Professional Certification
  • Geographic Information Systems Professional (GISP) · GIS Certification Institute (2012)

Go back

Your message has been sent