TotalEnergies Data Quality at Industrial Scale with Databricks

Summary

TotalEnergies runs more than 78,000 automated data quality checks across inspection records, maintenance systems, operational data, and industrial IoT sensors on the Databricks Data and AI platform to support critical barrier monitoring, where bad data could conceal the degradation of safety-critical equipment.
The architecture combines federated data products, Open Data Contract Standard contracts as a shared source of truth, SodaCL, DQX, Delta Lake, medallion architecture, and Databricks Unity Catalog to automate quality rules and tests across dozens of industrial sites operating 24 hours a day.
This video also covers Repsol's Area 2.0 transformation, which uses Databricks for governance, lineage, Genie, AI agents, and performance improvements as part of a strategy to close the gap between data-ready and AI-ready enterprise capabilities.

TotalEnergies Data Quality at Industrial Scale with Databricks

TotalEnergies is building data quality at industrial scale with Databricks to support process safety and critical barrier monitoring. The approach measures freshness, completeness and validity across inspection records, maintenance systems, operational data and industrial IoT sensors, where unreliable data could conceal the degradation of safety equipment.
Learn how TotalEnergies combines federated data products, Open Data Contract Standard contracts, SodaCL, DQX, Delta Lake, medallion architecture and Databricks Unity Catalog to run more than 78,000 quality checks. The session also covers Repsol's Area 2.0 transformation, including governance, lineage, Genie, AI agents, performance improvements and the enterprise context needed to become AI-driven.
Data quality management: https://www.databricks.com/discover/pages/data-quality-management
Customer data and AI use cases: https://www.databricks.com/blog/data-intelligence-action-100-data-and-ai-use-cases-databricks-customers

Chapters

FAQs

Why does TotalEnergies need data quality checks for process safety?

TotalEnergies operates dozens of industrial sites running 24 hours a day producing large volumes of operational and safety data. Critical barrier monitoring requires trusting the data behind safety equipment status, because unreliable data could conceal the degradation of safety systems before a failure occurs.

What tools does TotalEnergies use to enforce data quality on Databricks?

TotalEnergies combines Open Data Contract Standard contracts, SodaCL, and DQX to automate more than 78,000 data quality checks across its federated domain architecture. These tools work alongside Delta Lake, medallion architecture, and Databricks Unity Catalog to provide governed, lineage-tracked data quality enforcement.

What is an Open Data Contract Standard contract and how does TotalEnergies use it?

An Open Data Contract Standard (ODCS) contract is a versioned, machine-readable agreement that defines the expected structure, quality, and ownership of a data product. TotalEnergies uses ODCS contracts as a shared source of truth to automatically generate and enforce data quality rules and tests across their federated domains.

What is Repsol's Area 2.0 platform and how does it use Databricks?

Repsol's Area 2.0 is a data and AI transformation platform designed to close the gap between data-ready and AI-ready enterprise capabilities. It leverages Databricks for governance, lineage tracking, Genie, AI agents, and performance improvements as part of a broader DataOps and change management program.

Full transcript

[00:08] Good evening. Thank you for being here. So, my name is Nicola. I am head of data strategy for the digital for ACC program in Total Energies. And with me today Osama, data and AI technical leader also at Total Energies.
[00:24] So, for the first 20 the next 20 minutes, we'd like to share with you how we are building data quality I at industrial layer scale for use case that does not tolerate uncertainty, which is
[00:39] process safety. So, we'll tell you what we are doing today, what is working and what is still ahead of us. Let me start with about Total Energies for those who don't know us.
[01:03] So, Total Energies is a global multi-energy enterprise company around 1,000 100,000 people, more than 120 countries. We operate across upstream oil and gas, refining and chemicals, marketing and services, and increasingly
[01:21] in integrated power and renewables. What matters for us today talk is the operational reality behind this scale, dozens of industrial sites, everyone running 24-hour days,
[01:36] everyone producing huge volume of operational and safety data. That scale and the safety demands shapes every every choice we'll discuss for the next 20 minutes. Let me show you where the data and AI
[01:53] sits in our strategy. So, this is our strategic digital program in one picture. On the left, collecting and mastering data from all our industrial sites.
[02:08] Dozens of source, much more applications. In the center, AI is at the earth as the mechanism that turns data into action. And on the right, five business outcomes we expect AI and data to deliver.
[02:25] Safety, emissions, revenues, operational excellence, and saving cost and time. Today, we'll zoom in on one of those arrows, improve safety. Everything else we'll discuss
[02:41] flows from there. And specifically on the program we call digital for HSEQ. So, digital for HSEQ is structured around three ambitions. First, reduce major risk,
[02:56] process safety critical barriers, predictive failures of safety systems. Second, reduce human exposure, continuous monitoring of work sites, unmanned operations, and remote
[03:12] assistance. And third, reduce emissions, continuous monitoring and prediction of pollutant. All three are powered by the same set of data and digital enablers.
[03:27] So, that platform we can see at the bottom of the slides. Today, we'll focus on the first ambition, reducing major risk through critical barrier monitoring, and we'll show you what it takes data-wise to make that real.
[03:43] So, here is the question we want to put on the table. Can you trust the data being your critical safety barriers? Because process safety cannot tolerate uncertain data.
[03:59] A relief valve, an emergency shutdown system, a gas detector, each one is a critical barrier that people that protect people, the environment, and assets. Each one is monitored by data from dozens of sources.
[04:14] Inspection records, sensor feeds, maintenance schedules, operator entries. One bad data can mask a real degradation. The barrier looks healthy when it is not.
[04:31] That's why we are here today. Trust in the data on a use case like this is not nice. That's the whole point. Let's look at what that use what that use case actually looks like.
[04:46] So, critical barrier monitoring is in one sentence, a continuous health and health check physical safety barriers. Multi-source, the data come from inspection, integrity, maintenance, operations, and industrial IoT sensors.
[05:03] None of these systems were originally built to talk to the to the others. Multi-frequency, from real-time sensor readings to monthly inspection, different rhythm and are all relevant.
[05:19] And multi-quality, system fails without warning, human errors, system silently drifts, none of these failure announce themselves. So, what does trustworthy data mean concretely on this registry of critical
[05:36] barriers? Three dimension we enforce at every data product boundaries. Freshness, completeness, validity. Is the value in the physical and that's the point.
[05:57] Our approach has four ingredients. One, data domains, inspection, maintenance, operation, HSEQ. Each branch, each industrial site owns the data it produces. No central department speaking on behalf of the data they don't generate.
[06:13] Two, data products, an interface to the data, schema, semantic, owner contracts. Built once, consumed many times. On the diagram, you see different products, equipment, work order, downgrade situation, integrity status.
[06:30] Each one is a publishable, contracted assets. Three, one common data analytics platform, a common lakehouse foundation, federated by domain. No team reinvents the plumbing.
[06:45] And four, and this is the cultural ingredient, the journey is pulled by business requirements, not pushed by top-down data strategy. Each use case extend the foundation and each new products compounds.
[07:00] Each reuse what came before and it feeds what come next. So, that was the model. Now, I'll hand over to Osama who will take you through these
[07:16] three main things, technical foundations that make it real, the data contract that bind it together. Thank you, Nicola. So, a quick technical proof to demonstrate on how we are leveraging Total Energies data platform to serve
[07:32] the digital for HSEQ ambition. So, five pillars, one platform built around Databricks. First, the multi-cloud lakehouse. We are using compute and storage from different cloud providers.
[07:47] So, the idea here is that geography drives the platform, not the reverse. Uh we are using CI/CD pipelines. We are using asset bundles in for the deployment, Databricks workspace for the orchestration. Delta is used for the storage, the
[08:03] schema enforcement, and the evolution. Then, the medallion architecture. So, the classic bronze, silver, gold that you know. In Total Energy, we call it uh raw, curated domain. Maybe it's boring,
[08:19] but it works. And the key message here is that you don't promote data unless it's verified. Next, the governance, and it should be by design. We are using Unity Catalog as a source of truth for the access management, the lineage, the
[08:35] classification. And uh the governance should not be patched later, it's by design. The fourth element is the data quality and the maintenance, and sorry, the monitoring of the data quality. So, we are using mainly two framework. We are
[08:51] using Soda CL for the data quality on the data product, but we are also using DQx. DQx is Databricks data framework that is open source. The two data quality frameworks uh run side by side, but on different scopes. We'll have um more details later in this
[09:08] presentation. And finally, the data consumption and the data integration. So, the platform uh serves a diverse range of users from the digital uh sorry, from the data analysts, business analysts using reporting and
[09:24] dashboards. Also, business use cases, and finally, AI agents. So, the platform is providing standardized ways to consume the data. For instance, APIs, uh power platform integration, and so
[09:40] on, MCPs. And the one central element and one element that ties everything together is the data contract. So, what is the data contract? Data contract is the formal representation of
[09:55] the data product. Without a data contract, it's a concept, a slogan, but with a data contract, data product becomes an artifact that is versioned and deployable. Also, it's like we can see you can see it as a formal agreement between the
[10:11] data producers and the data consumers that outlines the uh expectations, the rules, and the responsibilities. So, everything that matters for the product, we can find it here. Um So, uh to go back to the uh reference
[10:28] example that Nicola talked about, the critical barrier management, so we uh from a point of view of consumer of the data, the critical barrier management and the data uh platform aligns on how to consume the data, the depth of the data, the frequency, the
[10:43] update, etc. And from a producer point of view, we are leveraging the data contracts and uh the data quality assessment to create MDMs, for instance, the critical barrier and the critical equipment referentials.
[11:00] We are using ODCS. ODCS is open data contract standard. It's an open source, vendor neutral, and backed by a large community. So, it's a specification. It uses YAML files, so it's a human
[11:15] readable and machine readable text files. And it contains all the specification of all the data contract. We'll uh the next slide we'll find the content of this uh data contract. Uh the data contract is the single source
[11:33] of truth and the central element for three audiences. The engineers that create the data pipeline and data quality checks, the stewards that validate the semantics based on the contract, and finally the AIs and the LLMs that use the contract as a ground
[11:49] truth. And the final secret ingredient is that this specification is open. So, it's we have the portability. So, we have different connectors backed by community. For instance, for the data quality, we can choose the engine that will write
[12:06] the actual tests. So, on this slide here, we'll see the content of the data contract. We'll talk about the implementation on the left side of the screen, and finally we'll talk about the results.
[12:22] So, the open data contract standards represents a data product across eight dimensions. These eight dimensions are what we see here in the screen, the ownerships, the terms, the models or the schema,
[12:38] the definition or the semantic layer, some examples, the quality rules, quality checks, SLAs, so the agreements, and the infrastructure. The contract is not a static file. It should be kept up to date when there are
[12:53] business updates or data source migrations or regulatory updates. Um, next for the implementation and the integration, so we are talking about hundreds of data contracts. So, we need to put in place
[13:09] an automatized way of um, converting the quality expectation from the YAML file to actual Python code that runs the tests. Um, also the to talk about the
[13:24] end-to-end framework of the data quality, the next step is to study data quality insights, the results, to create the dashboards, and to create um alerts based on the criticality of the quality checks. In these quality checks, we can find
[13:40] schema that is not matching, uh malformatted data, constraint constraints that are not respected, deviation when we talk about time series data. So, and finally, the results. Um so, today we are running more than
[13:56] 78,000 checks, and our quality score is above 85%. And the beauty of the data quality framework is that we see and we have uh more insights of uh on the um
[14:12] industrial process uh on the field. For example, the equipment, the critical equipment that are not tagged correctly or maintained correctly, an SAP maintenance that is running late, and these kind of things. So, the data quality conversation is not
[14:29] uh is no longer about the semester or the quarter cleanup. It's uh now a Monday morning conversation. So, I'll hand back to you, Nicola, for the close. Thank you, Souma. So, to conclude,
[14:45] three takeaways from what just we shared and what one commitment. So, one open standard plus data quality runtimes practical quality at scale on Databricks. ODCS for the contracts.
[15:00] So, the CL2 today, the QX tomorrow on Lakehouse native perimeters. Two engine, one contract standard. The choice of engine is reversible, the contract is permanent. Two, federated stewardship, accountability
[15:16] where the knowledge lives. Producers own uh the quality of their data, central plumbing uh, make that accountability cheap. And three, agentic AI is the next force multiplier, rule authoring
[15:33] triage which is failure classification, root cause analysis three activities that today are bottlenecks. We are experimenting with agents what can take that load. Early days, but the direction is clear.
[15:49] And one commitment, process safety is the forcing function. Data quality on a use case like this it is not nice to have, it is vital. So, much road remains. We need to bring DQX production on Lake Perimeter. We need to
[16:04] take AI agents from experiment to real assistance. So, that's it for today. Uh, if that resonates, don't hesitate to reach out. Uh, we still have two minutes if you have some questions.
[16:19] Thank you for your your attention. So, first of all, on behalf of Repsol, I want to say thank you to share our new uh, data and analytics project more than evolution that is starting to change our global uh,
[16:34] strategy. It's ambitious project that we have uh, take around a year and take the way to unblock the real value of of our data that we cite in in in our ecosystem. My name is Daniel. I'm the
[16:50] head I'm the global head of data and analytics in Repsol and I have uh, more than 20 years of experience to working with data and analytics. The last 10 years, I working in strategy and initiative and leadership of this kind of
[17:06] project. So, the idea is to help companies to become data driven, but in the case of Repsol, the challenge is more uh, exciting. So, uh, in Repsol, the idea is to become a AI driven company and this is the reason because we are starting to do this in incredible project.
[17:23] So, what is the the agenda? First, I starting to share our real challenge. How I how I try to align AI analytics ready and artificial intelligence ready, okay? Later, I try to explain the solution and how
[17:39] Databricks help us to perform this incredible transformation as a core component of the solution that unlock a lot of functionalities. Later, uh I issue that we have a conversation
[17:55] about business impact. Okay, what is the impact for the value of the business? And later or finally, we try to share our vision of how artificial intelligence is evolve in the long and short term.
[18:16] Before I start with the challenge, let me take a minute to share or or starting to employ to show the company, okay? About Repsol. Repsol is a multi-energy company like Total. We are operating in more than 20 countries. We have more than 24 employees to support more than 20
[18:33] millions of customer. Part of this customer is digital customer and is very important for us because Wallet is our digital app and we are capable to tracking and personalize all of the offers that we are capable to offer for our
[18:50] digital business user. Repsol has four tradition, sorry, four business. The first is upstream, is our traditional business, okay? The second is industrial. We have seven industrial plants and plus one fuel renewable
[19:08] plant, okay? Later, we have a renewable business because we try to comply with our objective of to be a net zero company for 2050. And clients or retail, this is more than
[19:24] or or the the point, thank you for the time, the point for a B2B and B2C strategy happen, okay? Everything related with service station, gas, electricity happens here in our client direction, and we have in this moment more than 3 million of customer
[19:41] with gas and electricity. And last but not least, innovation and technology the the corporate area. This is the part that we are developing our business digital program, okay? We have two ways of the business digital program and unblock more than 1 billion of economic
[19:58] impact to the company. So, this is the area that we are working in this moment with a lot of digital business cases. In this moment, more than 1,300. Before to start the challenge, let me to push as Ali said in the keynote, context,
[20:16] okay? Context is very important. So, this is our timeline or this is the reason because we are building everything. In 2018, we starting with our business digital program, as I said. And the data is starting to grow in fast, okay? We starting to unblock a lot
[20:33] of digital initiative cases. And the point is is that we need a data and analytics platform. So, this is the reason with we build area. Area is our data data and analytics platform and we starting the travel between 2021
[20:50] and and build our data warehouse as naming data driven product is the big a product that we have. And during 2021 until 2024, we are or or establishes our data driven foundation, but what happened later? The last two
[21:07] years, mean this is here last apparatus did. I just starting to work in resolved two years ago. We starting to thinking in how we are capable to unblock and scale more quickly the value of the data and and the solution for our
[21:23] digital user, okay? So, this is the reason because we build area docs zero. So, with this name you imagine that I am working in in data and analytics and not in marketing. Our new platform is similar to area but with a lot of new functionalities, okay?
[21:43] This trouble is starting with a personal reflection from my side. Two years ago, we starting here and explaining how we are capable to build LMM machine learning lab, okay? Now is our artificial intelligence lab. But we have a a challenge or or a room to improve me
[22:00] here because if we are capable to have an incredible level of maturity machine learning and in and AI will happen with data. What's going on here? This is a tremendous reflection for this ambitious process project, sorry. So, in
[22:16] this moment when we starting analysis, Harvard said that the chief data officer has two direction, okay? Offensive and defensive inside the strategy. So, for the offensive parts, yes for the offensive parts, sorry, the performance
[22:31] and the start to date and the time to data is starting to have a lot of problem because the scale down scale down and scale up computer resources is not enough for a lot of digital business cases, uh specifically for the medium cases,
[22:48] okay? So, we have a room to improve because we are starting to lose adoption with the data product. Number second, data ops. We have data science teams and data intelligence or business intelligence teams working together, but not in the same speed.
[23:05] It's not synchronized and it's one of the problem or the reason because we starting to develop area 2.0. For the defensive part, we have a incredible pain point with with cost control. Why?
[23:22] Because we have a lot of technical debt, okay? We have lot of observability. We don't have a a clear or easy way to see of our data engineering process is scaling or not is scaling. So, the point is that
[23:39] the technical debt is starting to reduce the benefit of our digital business cases. So, we have a problem here and with the data governance, specifically with metadata and lineage conversation. Why? Because in metadata, our row level
[23:54] security is not easy, okay? We couldn't have for the same KPI to the different person and obtain different result. And it this is very, very important for now, for Genie, for natural conversation. We need this kind of data governance tips.
[24:09] So, we starting to change it, the direction and of course, the lineage in the data governance process, we don't have a end-to-end lineage problem. So, we have a room to improve here because for maintain all of the data governance of the platform, let me introduce later, we
[24:27] have a lot of manually work to do. So, at least innovation, but of course, in Repsol, we have a company that evolve and and and ambitious to be a data driven company. So, we need a partner to work
[24:44] at the same innovation speed as us. But I don't see anything a lot of because in the keynote, we receive a lot of updates. So, clearly, uh is our partner and have the innovation that we needed in this moment.
[25:01] So, when we are make this kind of assessment that I showed you, we have we we have deeper in the problem. So, the point is that the gap between AI ready and data ready is hidden, okay? But but starting with a
[25:18] hidden cost, the decision are made too low. And we have a business problem here. So, the idea is that everybody feels to work aligned, but everybody is starting to work in different streams. So, if you see the business intelligence
[25:34] tools on way on the left side and the AI or machine learning data science tools is very very different. So, the three big gaps is we are AI ready, but not decision ready because our business intelligence cycle is too too long. We
[25:50] have two different way of working and culture, okay? And the platform is not a typical base camp to work and to share. We have a problem here because it's starting a piece of friction because data science work in a way and business intelligence teams work in another
[26:06] different way. And for end, the data lake layer is not aligned, okay? We have different layers for different use cases and the only common layer is the raw data, bronze, okay? If you call
[26:21] it bronze, so um is a big challenge and this is later for the business impact, but let me to bring the one North Star case is so we have a location intelligence problem
[26:36] that we are happening one of our industrial complex. And for example, when we are starting to unify everything in Databricks, we are capable to faster processing, reduce the cost, and of course we obtain a time saving. we have
[26:53] 19 data driven in our company, 19 data driven project big product. Along with this product, we are starting to see the benefits of working together and in a live way. So, to continue with the presentation, let
[27:09] me explain how Databricks is a core partner and the key part of our solution. On the left side, we have Area, the first version of Area, that uh support more than 500 uh use cases. This is
[27:26] another version, 500 uh new cases. Uh we have more than 400 source system integrated in the platform. And in the left side, the four pillar of Area is data and security access, data governance and data quality, data ops
[27:44] because business intelligent teams work uh with a BI framework. And the self-service BI to unblock the delivery program problems, okay? But in the second version of Area, the new our data and analytics platform, we add
[27:59] Databricks and we are capable to have, for example, data governance, the lineage that we are talking about. We have a end-to-end lineage now and we are capable to see what happened with all of the objects in Area. We are starting to work in a synchronous
[28:15] way because we have a AI ops framework. Use data federated Data federated is one advantage here in Databricks. Why? Because our data governance model is federated, too. So, we are starting to explore a lot of insight about how we
[28:31] are capable to data to govern on data that we don't have in the platform. And it's a variable a incredible way from this part. So, the final two axis of the new platform is generative BI, like Genie. Why? Because natural language
[28:48] conversation is the future. So, everybody is starting to have insight in natural language body in natural language, sorry. And the point is that dashboard is not the end of the conversation. In this moment, we are starting to increase and tested and we have the the
[29:06] first axis wing cases. So, second the agent. In this moment and I suggested to review the Victor tall of of this morning because we exploring agent with Genie code, with agent
[29:21] briefs, with the knowledge assistant. So, this is very very important for us and we are starting to explore and work a future workflow identification to work in in the platform. And of course, sorry, and of course, everything with the boring as
[29:38] total said, the boring architecture of medallion, okay? But is very important for us because in the past, we have more than three more than three layers in the data lake and reduce all of the speed in the delivery. So, in
[29:53] this moment, we are working with the the medallion architecture. So, let me share three tips to working in this kind of big big project. The first is the strategy. We need to the empowerment of the board level, okay? So, in our case, is not a problem
[30:09] because our CEO is an advocate of the digital initiative program. So, for this for the inside, we have everything that we need. Later, we need to collect the transformation, okay? We are enabled but the real leadership of the initiative is the
[30:25] business director. This is the only way to increase the adoption. Later, we are working in a similar way in digital and IT department, okay? We are better together. The key partner, of course, we are here today in the room. So, I appreciate Databricks and Cedal
[30:42] to work here with us. Databricks for putting a platform that innovate with us at the same speed and and unblock a lot of functionalities. And Sidetrade very important because it's not a only a service consultant.
[30:57] Okay, no. It's a co-architect of this transformation. This is the example what we build together. The archetype is our new AIops framework with different flavors and everybody capable to work in the same way. Very important for us. And
[31:13] change management because this project is very very big and we have more than 40 professional working in different areas with different times with different cultures. So, we are managing and high performing teams. Okay, so my tip here is explain the goal and working
[31:30] together and try to manage everything. In waterfall, in agile, is not important. The important for us is put the objective and advance at the same speed. The most important part of this conversation, the impact measure. Okay,
[31:46] when we starting this conversation, everybody starting to to say, "Of course, Databricks. Wait Wait Wait a minute, Danny. So, what is the business impact for the for the What is the business measure for the impact?
[32:01] Okay? So, from the technical perspective, we are migrated around 34 billions of record. Okay, in this transformation and we are capable to improve 57% the average performance. So, it's a
[32:16] incredible advance for our adoption. Another point that I'm to highlight is we are capable to reduce the line of code. Okay, we have more than 450 millions of row and we are capable
[32:32] to reduce to two 276 million. And of course, with this part we are starting to increase the adoption. We are capable to increase the monthly user close to 40% after this transformation. So, we have
[32:47] the first um with wins for the technical overview and the business impact, let me rescue two cases. First is a customer intelligent. Remember Waylet, our digital application. We are capable in the new platform for tuning
[33:05] all of our loyalty models, okay? I'm customizing the workload very, very quickly. Before this project, we need um a lot of hour to analyze the capability of of the loyalty customization
[33:20] workload. And later, we are capable to be more faster. And second, GLP operation. We have a lot of decision to do with cylinder gas before 8:00 a.m. So, with the new platform we are capable to put every day in the same place, and
[33:37] we are capable to reduce the energy savings and increase the productivity gain. So, let me to share before finish what is our future vision about artificial intelligence. Sorry because in Repsol we
[33:52] are believe that artificial intelligence transformated and change everything. So, first of all, this is the linear process decision that we have 2 years ago in Repsol, but if you see this project this
[34:08] Sorry, this pro- process decision is the same for a lot of company. You take your soul, move the data, generate insight, and create action. During the last year, we are working and analyze how about we are capable to a scaling.
[34:23] How we are capable to resolve the ontology program, and very interesting the new advance of the new ontologies today. We have thinking about how we are capable to manage the inside driving conversation, the new knowledge management, and who we
[34:39] are capable to advance with analytics capabilities. So, now, this is an image of my personal genie. We are capable to a scale and increase all of the all of the digital transformation asset
[34:55] and the service that we are capable to allow from the platform. And in this district, the thing that we are learning is with the right context, agent is capable to take better decision and improve and help us to optimize the process. Without
[35:12] context, it's only an assistant. So, we are build this process to make a incredible investment to build our new enterprise contest layer without technical limitation. So, we are very very happy for this part, and this is my
[35:29] final reflection. So, for the next year, Gartner said that the 40% of agent maybe shut down for performance cost related to token war to token war or um data governance perspective. So, the
[35:45] question in Repsol is not how many agents we are or will be shut down. The question in Repsol is how we are capable to do with this competitive advantage. Thank you very much.

Learn more about the Databricks Data and AI platform.

The information provided herein is for general informational purposes only and may not reflect the most current product capabilities or configurations.