A quality engineer who learnt to distrust a record nobody could assemble

The first data problem I owned was a furnace chart. In the spring of 1999, ten months into my first job as a quality engineer at Ferronde Forge in Trois-Rivières, an aerospace customer rejected a shipment of forged titanium rings because the circular chart recorder on one of our heat-treatment furnaces had run dry for forty minutes of a cycle. The rings were very probably sound. Nobody could prove it, so they were scrapped: C$212,000 of material and four weeks lost on an engine programme. I spent the following month with the furnace operators working out why the chart mattered more than the metal, and I left with a view I still hold. In our industry a part is only as good as the record that travels with it.

I grew up in Drummondville, the daughter of a millwright and a school secretary, and studied industrial engineering at Université de Haute-Rive, finishing in 1998. Ferronde made closed-die forgings for aircraft engines, rail bogies and hydroelectric turbines, and it gave a new engineer responsibility quickly. By 2001 I was running statistical process control for the forging presses. That taught me that operators trust a control chart only when they have helped decide what goes on it.

In 2003 I moved to Castelmont Précision in Longueuil, a privately owned machine shop of about 380 people making rotating parts for engine manufacturers. I led metrology and quality data and later the quality system. Most of our records were paper: travellers signed at each operation, inspection reports, coordinate-measuring printouts stapled to the back. I built a database for first-article inspection reports and nonconformances, and studied applied statistics part-time at Université Mont-Ferrier, completing a master's degree in 2006. When a US group bought Castelmont in 2008 and moved its quality systems to its own head office, I chose to leave rather than administer a system designed for somebody else's plants.

My husband's work took our family to Sudbury that year, and I joined Brantwick Mining Equipment as manager of manufacturing systems and field data. Brantwick built underground loaders and drill rigs, and its machines sent telemetry from mines in northern Ontario, Quebec and Chile. I was asked to predict component failures for the service business. I learnt how little a sensor stream tells you without a reliable record of what actually broke and why. Our warranty claims described failures in free text written by dealers who were paid per claim, and the first models we built predicted the writing habits of dealers better than they predicted failing hydraulic pumps. We rebuilt the failure coding with the service technicians before we built anything else. Records before models has governed my work since.

Tessandier recruited me in February 2014 as director of quality systems and product traceability, after an engine customer's audit found that some of our special-process records could not be tied reliably to the parts they described. I spent four years on that problem: one nonconformance system across the plants, electronic furnace records at Boucherville and the first serial-number links between the plants' manufacturing execution systems. In April 2018 I became vice-president of digital manufacturing and operations data, responsible for plant systems and the analytics team. In January 2022 our chief executive, Benoît Lavallière, created the role of chief data officer and asked me to take it.

Since then my work has turned on four questions. Can we tell a customer, within hours, every part made from a suspect lot of material? Which maintenance and quality decisions should a model inform, and which must stay with a certified person? Who owns the data a plant produces when head office wants to compare plants? How do we use data about inspectors and machine operators in ways that respect Quebec's privacy law and keep their trust? I have answered each imperfectly, and the cases on this site record what the answers cost.

I work through the people who make and inspect parts. I spend at least two days a month in a plant, usually with maintenance planners, inspectors or the plant data stewards we fund jointly with plant managers. I ask for the failure history before I look at a model and for the paper traveller before I look at the dashboard built from it. Colleagues would say I am slow to announce a programme, quick to publish a definition and stubborn about writing down what a system cannot do. I would rather tell the audit and risk committee that a drill took three hours and found two errors than show it a green indicator nobody has tested.

I work in French and English, and in Spanish well enough to walk the floor at Querétaro with its general manager. Outside Tessandier I sit on the board of Atelier-école Rive-Sud, a not-for-profit that trains machinists on Montreal's South Shore, where I learnt what a director can and cannot see from a board pack.

Open to Board appointments · Advisory roles · Speaking invitations

Languages / French · English · Spanish (working knowledge)