Total Energies

TotalEnergies is a giant. Founded in 1924, they are a global powerhouse with over 100,000 employees across 130 countries. We’re talking the whole shebang: oil, biofuel, natural gas, green gases, renewables, and electricity.

But here’s the thing about legacy: it can be heavy. TotalEnergies didn't just want to rely on their history; they wanted to innovate like a startup to withstand the test of time. They wanted to use data to tackle climate change and make energy cleaner, reliable, and accessible.

TotalEnergies had a vision that would make any data scientist excited: they wanted to build predictive models around usage data, weather patterns, and demand forecasting.

But their internal tech stack was hitting a wall.

• The Reality: Their Business Analyst (BA) teams were sitting on a decent infrastructure, but it wasn't scalable.

• The Pain: Systems took too long to run, parallel workloads were impossible, and the setup was frustratingly slow.

They wanted to be efficient. They wanted to be adventurous. And frankly, those are two qualities we bloody love at Razor.

We don't believe in guessing. We believe in testing.

We kicked off with a discovery session to understand the true gripes of their existing process. We knew we could build something awesome, but we needed to find the right tool for the job.

We pitted two heavyweights against each other: Azure Databricks vs. Azure Machine Learning.

We didn't just write a report (though we did write a pretty awesome pros and cons list). We went a step further. We built simulations with example inputs on both platforms, letting TotalEnergies test drive their future architecture.

The Verdict? Databricks took the gold. It offered the room to grow, the scalability they craved, and the capability to handle everything they needed, and then some.

We built a cloud-based powerhouse that transformed how TotalEnergies handles data.

• Custom Python Models: We created a solution allowing them to build their own Python models to track data provenance with precision.

• Embracing MLOps: We moved them toward Machine Learning Ops (MLOps). This isn't just buzzword bingo; it allowed their teams to collaboratively build and run data models on the cloud using massive compute power.

• Scalability: No more bottlenecks. They can now run parallel workloads and scale as high as their data takes them.

TotalEnergies is now fully set up on Databricks, running complex Machine Learning workflows in the cloud.

They moved from a system that was "frustratingly slow" to a collaborative, high-speed environment that lets them predict the future of energy usage. They have the tech to match their massive aspirations. Wowsers.