Context and assignment
Data science for energy performance and predictive maintenance.
Optimisation of energy performance, predictive maintenance and data normalisation. Failure-prediction algorithm to anticipate maintenance, mobile reporting app with voice recognition.
The challenge
What was at stake
For Engie, the assignment covered: “Data science for energy performance and predictive maintenance”. This kind of programme involves business departments, IT and data teams at the same time: it has to deliver quickly without weakening existing systems, and leave behind practices that hold once the assignment ends.
The relationship
Visian since 2020
ENGIE is Visian's longest-standing and most diversified client: 20 group entities, from corporate to field operations. Visian has worked there since 2020, with 96 assignments across 20 group entities.
Expertise involved: Product management, Databricks and Spark data engineering, full-stack development, governance (data steward), AI change management.
Deliverables
- Product management and data engineering for MaConso (energy consumption tracking)
- Global service centre
- Databricks architecture and data squad
- Expert data steward
- Copilot change management
Team and duration
Duration, team and scope
- Duration: since 2020.
- Team: a multidisciplinary Visian team, sized to each phase, covering product management, Databricks and Spark data engineering, full-stack development, governance (data steward), AI change management.
- Scope: Product management and data engineering for MaConso (energy consumption tracking); Global service centre; Databricks architecture and data squad; Expert data steward; Copilot change management.
Approach
How Visian worked
- Scope: workshops with the departments involved, assessment of the existing set-up, prioritisation and deliverables approved in steering committee.
- Build: delivery in short iterations, shared milestones and project governance with the client teams.
- Handover: adoption support, documentation, skills transfer to in-house teams and move to run.
Results
Measured results
Scope
Methods and technologies
Expertise involved