How Amadeus Modernized aviation operations on Databricks

Summary

Amadeus migrated its aviation analytics platform from Azure Synapse to Databricks in three months, unifying operational data across three geographies using Delta Lake, medallion architecture, and automated pipelines.
The modernization enabled Amadeus Max, an AI-powered operations advisor that answers complex queries in natural language, reducing manual reporting time from weeks to minutes for airlines and airport operators.
The migration delivered 35% cost savings and was completed 50% faster than a standard timeline by leveraging accelerators and brick builder solutions to handle differing schemas and refresh cadences across regions.

How Amadeus Modernized aviation operations on Databricks

Aviation operations are fragmented across 400 plus airlines and 200 plus airports, with passenger journeys disrupted by delayed visibility and reactive responses. Amadeus needed to consolidate silos of operational data from booking systems, baggage reconciliation, and gate assignments into a unified, governed platform capable of serving real-time intelligence without compromising geographic compliance.
See how Amadeus migrated from Azure Synapse to Databricks in three months, achieving a unified data foundation across three geographies with Delta Lake, medallion architecture, and automated pipelines. This modernization enabled Amadeus Max, an AI-powered operations advisor that answers complex queries in natural language, reducing manual reporting from weeks to minutes. Learn the data strategies, accelerators, and patterns that delivered 35% cost savings and positioned the platform for agentic intelligence.
🤝

Chapters

FAQs

How did Amadeus migrate from Azure Synapse to Databricks in just three months?

Amadeus used Databricks accelerators and brick builder solutions to handle the complexity of unifying data with different schemas and refresh cadences across three geographies. The team made hard calls about what to standardize, what to preserve, and what to let go to hit the timeline.

What is Amadeus Max and how does it help aviation operations?

Amadeus Max is an AI-powered operations advisor built on the Databricks platform that answers complex operational queries in natural language. It reduced manual reporting from weeks to minutes for stakeholders across airlines and airport operations.

What compliance requirements did Amadeus have to meet during the migration?

Amadeus had contractual mandates requiring that no geography could see another region's operational data. This was enforced as a hard requirement in the unified platform design, meaning geographic data isolation was treated as a compliance obligation, not just a technical preference.

What legacy problems did Amadeus solve by moving to Databricks?

Amadeus's prior platform on Azure Synapse worked in silos and lacked the efficiency needed for real-time operational intelligence. Fragmented data from booking systems, baggage reconciliation, and gate assignments made it impossible to deliver proactive responses to disruptions across 400+ airlines.

Full transcript

[00:07] Hello everyone, good morning. I'm Abhishek Krishna. I lead the data and AI product team for our airlines and airport operations portfolio. I've been in the aviation industry for over a decade, started off my career as a software engineer for for for an airline and eventually transitioned as a
[00:24] product manager within Amadeus. Hello everyone. I'm Gurpreet Arora. I lead data and AI for travel and hospitality at Tridant. Um in the last decade my career itself has been a journey. Started as a data scientist. Um
[00:40] I'm now working with Fortune 500 leaders across the globe helping them realize the value of data and AI. Um Personally very passionate about travel and fortunate to be working in the industry. Um and that gets me energized to share one of these um
[00:57] stories today that how we modernized and we actually enabled agent AI for aviation platform operations with Amadeus. Uh excited to share the story with you guys. Um hope it's a good one for you.
[01:17] All right. So, I wanted to start with the good stuff up front, right? Uh this is the so what slide. I could talk about the architectures all day. I but this is what actually moved the needle for Amadeus. I want to start with what we actually built
[01:34] and what did it deliver? So, Amadeus has an analytics platform which was initially on Synapse. It worked in silos. It did a lot of things, but it was not the most efficient. The kind of efficiency that you need for something like operations
[01:49] which I'm sure you guys have experience in different setups. Um what we achieved is we were able to unify global domains, three geographies in three months. Right? When I say that the it's
[02:04] different regions, the data for these regions was just not in different locations. It had different schemas. It had different refresh cadences. Um consolidating that into a single governed platform in three months meant we had to make some really hard
[02:20] calls uh what to standardize, what to preserve, what to let go. And no one sees uh the geography in another region's operational data because this is not just a technical requirement. That's a contractual mandate by Amadeus. Similarly, we actually I already spoke
[02:38] about the speed. I wanted to double click on we were able to do this uh 50% faster timeline and I will give a double click on how did we really achieve that, right? Through our accelerators and brick builder solutions. Um we were able to augment and empower
[02:53] agent AI. Uh operations advisor is such a critical aspect given the number of stakeholders involved in resolving resolving a disruption setup. Uh what this modernization of plat- platform really enabled is start that agent AI intervention, start
[03:10] building those agents on top of this platform as it's been the theme throughout the Databricks Summit that your agents are only as good as the data. Once we had the operational side and the uh platform side figured out, we really made sure that these agents are efficient and doing the job that's
[03:27] required in a critical setup for disruptions. Um total cost of ownership, this is one of our favorite metric cuz sure we hear the stories about migrate, modernize uh everywhere, but what is the true cost of ownership? That includes your one-time migration cost, that includes
[03:42] how well did you engineer and optimize? And this is a real report card for uh us to really see if the migration going from Azure to uh Synapse to Databricks, how efficient was it? And we are able to track 35% uh saving post migration.
[04:04] Talk about the operational agility. So, uh airline ecosystem are highly fragmented. Well, most of the travel ecosystems are. Uh you have PSS, you have baggage reconciliation, bag crew scheduling, rostering, flight schedules. So, we really had to create this unification of a lot of operational
[04:19] aspects underneath so that these interactions and agents can actually work on this single source of truth. And this line has been in business for more than 10 years, right? So, shifted to predictive. Uh most of the operations today,
[04:36] uh it's a dirty secret in uh airline industry that it happens over walkie-talkies or WhatsApp groups or, you know, we see people running around and trying to handle disruption. Can we anticipate and have a more system-based approach where the
[04:53] alerts are proactive? We are not really hampering the brand reputation or the customer experience. So, this speaks to it. So, this is what was delivered and I wanted to just put it right up there. What you will see in the following slides is a little double click on uh what we have What was the problem
[05:09] statement? What is the entire scope? What is the use case we are really solving, which I'm sure you guys will resonate with and a little little bit on the architecture side. I'll start with the partnership and a quick introduction for Tridens. So, I'm from Tridens. Uh Tridens is a global AI
[05:25] and data science partner focused on solving the last mile AI adoption prob- uh problem. Since our inception in 2013, uh we've been really focused on delivering the last mile AI adoption and do it faster. Uh we have partnerships with hyperscalers
[05:42] and cloud providers like Databricks. We have brick builder solution accelerator for this platform switch help us deliver value at speed. So, we typically do this 30 to 50% faster than most of the SI's in the space. Uh
[05:58] We have also focused on a lot of technical and domain accelerators. So, it's not just technology accelerators. In travel and hospitality, we have accelerators like unified data models. We have focus for hospitality, airline, air ops, which really help us realize this value at
[06:14] speed at scale, something that we did for Amadeus as well. As you can see, we are six times partner of the year with Databricks. Uh This year we got the business transformation partner of the year. So, shout out to my team here. Great job. Uh I've been personally part of Tridens uh and I've seen it grow. I was the 65th
[06:30] employee. Now we are 4,500. Uh it's just growing in last 8 years. Uh and as of now, we have more than 150 live Databricks use cases across 86 customers. Uh We just keep the growth going. Uh with that being said,
[06:45] I'll hand it over to Abhishek. So, um travel is operationally complex. I think uh we have a few airlines and airports in the room as well. And you're quite aware of the complexities involved, the multiple stakeholders who are part of the ecosystem.
[07:01] Uh at any given touch point, our solutions touch over 400 airlines, over 400 airports, 150 ground operators across uh 190 countries. And the This shows that we have extraordinarily high volumes of data that we deal with. Everything to do with
[07:17] passenger movements, baggage journeys, disruption signals. And historically, these assets were always uh always living silos. Uh and they were really accessible to a few specialists.
[07:34] So, overall, within our airport and airline operations vertical, we have around 210 uh airport operators, 95 ground handlers, over 150 airlines. And over 30 plus
[07:49] border control border authorities in the ecosystem. Our main purpose is to provide seamless journeys end to end and to optimize and make aviation operations more efficient.
[08:05] Our foundations is on keeping the platform open with the multi-cloud technology. We are not just a software company. We have acquired We have had some acquisitions in the past where we acquired
[08:20] hardware companies who build kiosks and backdrops. So today we have a manufacturing and a production line for our self-service kiosks kiosks or backdrops or biometric solutions. We have a unified data foundation to enable collaboration as you know in
[08:36] an operation ecosystem. It's highly complex, the multiple systems, multiple processes, multiple rhythms in the way the data is exchanged. And collaboration is always a challenge. Our solutions within Amadeus is pre-integrated end to end.
[08:53] And the main objective is of course to achieve end to end passenger processing and end to end operations overall.
[09:08] So we are in a very unique position within the portfolio because we have a combined suite of applications from airlines and airport operations. And this is where this is how we connect the dots. That's one of the reasons why we wanted to unify the the platform. So our portfolio
[09:23] has three main value streams. The first one is seamless travel. And in seamless travel it's an integrated suite of solutions comprising of user-friendly hardware like this self-service kiosk, the backdrops, the touchpoints as part of
[09:39] the passenger journey. We have digital identity and our innovative biometric capabilities. And then the second value stream is on delivery. So, our delivery management is the next generation of airline systems. Our delivery systems will gradually
[09:54] replace our departure control systems. And for those who are not aware of departure control systems, they are the solutions that are used for processing passengers either through mobile check-ins or at the airport. We are moving towards the new form
[10:09] retailing, the offer and order world. We are moving away from standard way of managing PNRs. And once the offer and order is created, we need to deliver it. And with our DMS solution, our main objective is to help hyper-personalization
[10:25] as part of the delivery process. And finally, we have unified operations. And within this suite, we basically cover all products within the airside. Every Every product For example, managing the airport operations plan,
[10:41] managing the fixed assets, managing the runway capacity, maximizing runway capacity. And also offer solutions such as the flight control solutions and the hub management solutions for airlines. And the foundations of it is to
[10:56] have an open data platform, and that's what we did with our recent migration to Databricks within the operations space. So, before we jump in, I think to add more
[11:12] context, let's make this a bit more real, right? Let me introduce you to Alex. He's a frequent traveler. He has a Monday morning flight at 6:30 flying from London Heathrow to Dublin. He has an important client meeting. So, there's no margin for error. He has
[11:28] to reach on time. So, quick question. How many of you have done this journey? Okay, good because what happens next has probably even happened to you. Boom.
[11:47] There's a weather event. So, Alex arrive at the airport and only then finds out that the flight is delayed. He queues queues for on 40 minutes uh to speak to someone who has the same information that he has. And then, guess what? The flight is cancelled.
[12:03] There's complete chaos for rebooking. Alex is uh booked on a flight much later. And uh once he lands, uh his day gets even better. His his bag disappears. So, there's no
[12:20] tracking, no updates, no accountability. This is not a bad day. This is a normal day now in in aviation operations. In a year, we have around 36.5 million missed uh mishandled luggage. And our
[12:38] disruption costs go up to 60 uh 60 billion dollars per year. That's a big number. Now, let's have the same journey, same passenger, but powered by intelligence.
[12:59] So, uh Alex, he gets a notification around 3:40. He receives Okay, his flight is slightly delayed. Uh and a message saying that they already found alternatives. Before leaving home, he taps once, rebooks in seconds. The system understands the context,
[13:15] understands that he has a meeting, there's an urgency, and provides the best rebooking option available. Alex gets rebooked in the right flight, arrives, his bag is tracked end-to-end, and he makes it to the meeting.
[13:34] So, when you look at today's reality versus the AI-enabled potential, there's this gap that we we must fix and fill. And this is an opportunity gap. When you look at disruptions today, you're not notified proactively.
[13:50] Whereas from an AI-enabled potential, we can in Alex's case, the weather data was augmented with operational data and based on those assets and connecting the dots, the information was sent over to the upstream systems either customer-facing applications or to the operation control
[14:07] center and the passenger recovery center so that they can provide the right options for Alex. And then you have the bag tracking visibility in today in today's uh journey, I think there are a lot of uh data gaps around baggage journey. It
[14:22] uh as part of any baggage journey, you have multiple uh owners who who manage the entire process. But with tracking visibility in the future, we're looking at connecting the dots with RFID, with tracking, with e-tags and tiles um and and having using computer vision.
[14:37] For instance, uh today in our auto bag drop self-service uh products, we take an image of the bag as as soon as it's checked in. So, we can re- and reuse and refer the image to to to manage the end-to-end uh tracking process. And then the self-service rebooking
[14:53] success in Alex's case didn't work out, but now we're trialing uh with our intelligence solutions around self-service rebooking based on recommendations. We've already seen in a trial around 40% um uh reduction uh in and call center
[15:08] calls. So, with that being said, uh let me pivot it to what a modern traveler really expects and where do these legacy operation platforms struggle to deliver?
[15:24] I assume all of you qualify to be a modern traveler, right? What are our expectations we have through Alex's journey that Abhishek just talked about? It's a chaos, right? Uh So, I just wanted to highlight few points and connect it to
[15:41] what is happening on the platform side and what's required and how we can we have rather uh made an attempt to close that gap. First thing is speed at scale, right? So, 5.2 billion passengers is an abstraction. What's real is
[15:57] agent at the Heathrow Airport with 40 misconnected passengers and 6 minutes before the door close, system is still showing round flight is on time. That's the gap we're talking about, right? Every second matter when the alternative is you have to stand in a queue for a
[16:13] voucher. Guilty of doing that. Uh second thing is always-on reliability. Like I said, the dirty secret in industries most disruptions are discovered through tweets and on social media. Uh not a consistent system alert. Some
[16:29] systems do catch it, but it's not a consistent flow of uh alerts that we see. A crew timing out is predictable 90 minutes before, so data exist. What we really need to do is build that layer and act before the damage is done.
[16:45] Third thing being the real-time truth. We had a situation where the bag is shown uh on belt and in transit into systems at the same time, right? That's really not a technology failure. That's a mistrust in data. And like we keep
[17:02] hearing about, we can do all this amazing stuff that Databricks has been releasing on Genie ontology, but it is all good if your data is actually consistent through the system. And that foundation aspect is where we really really have to focus given the diversity of the systems and
[17:19] the data sets across geographies that we were trying to combine. Connected operations, I think seamless coordination across airline, airports, ground staff, crew is extremely important. And if you don't
[17:35] believe me, go to an op center and see how the disruptions are managed. They are happening on WhatsApp chats, walkie-talkies. In some cases, the systems do catch it. It's not consistent throughout every stakeholder that is involved in addressing a disruption.
[17:51] And intelligent automation. We've had this case where a $15,000 customer lifetime value member is being treated the same way as a first-time flyer, right? Just because the systems were did not connect with the loyalty and the customer inferences
[18:07] and data set in the back end to make sure that what is the personalized recommendation or a rebooking or a voucher that they can enable. Even lounge access in some cases. So, having this streamlining the foundation of all your
[18:22] operational plus your customer flight schedule operations data sources really help you to deliver that personalized experience, especially when you're servicing a disruption.
[18:38] We leverage Databricks to build the solution. And I'll just give a quick overview of This slide is really about one thing, the gap between data existing and data being useful, right? Every problem we solved here was some version of that gap. And I can take you through some of the points.
[18:54] First point being the lag reduction from information to action. Okay, it shows now. Um So, like I said, these detections have been in different systems in fragmentation. How can we actually reduce the lag?
[19:10] That's where we did some engineering effort. We enabled the observability from a single hub. This is an incredibly difficult thing to do. And I'm sure peers here from different airlines and airport operations will resonate with this. Um before this, the operation team was
[19:26] looking at four to five different systems to understand what was happening on the ground at any given moment. PSS data is in one place, baggage is in another, gate assignment is in a third. Um third thing is we really wanted to curtail some redundancies in operational processes,
[19:43] right? This one is less glamorous, but possibly the most impactful from reliability standpoint. Uh the legacy architecture had data making unnecessary hops. It would land somewhere, get processed somewhere, get moved and processed and get copied again in different systems.
[19:59] Uh this is where we spent a lot of time. Trust me, through our engineering, acceleration, and whatnot, before we can bring you the goodness of latest and greatest that Databricks has to offer. And the last one being empower agentic intelligence. Of course,
[20:14] the whole point of making sure that it's unified and operational is can I just ask my advisor chatbot where are what are the node congestions on my airport? Uh where am I seeing the maximum baggage is being tracked or lost in the system?
[20:29] Can I combine the passenger and the trip uh history to make more sense of how do I service this uh passenger from a misconnected flight? And I'll go to the architecture side. So, this is for all the
[20:45] geeks and tech savvies in the in the room. I see a lot of cameras popping up. Uh So, we engineered Databricks services to actually meet the scalability and execution rigor. What you're looking at is the end state. Let me tell you what happened and what it replaced and why the replacement
[21:01] wasn't straightforward. First reason being, Amadeus is not a startup. As Abhishek explained about the complexity and number of airlines and operations they have, this platform serves airline airports globally. So, when we talk about migration, we're
[21:16] not talking about moving a data warehouse. We're talking about replacing pretty much the backbone of entire operation. Let me go from left to right. The data sources that you see, notice how many systems are feeding this platform. We have ACUS, we have baggage
[21:31] reconciliation system, we have APV, self-service kiosk, biometrics, uh ASM, all of these different sources. Each of these had their own schema, ingestion cadence, own failure mode, own notification.
[21:47] In Synapse, uh each one had a crafted pipeline. We had almost 15 code repositories across, like I said, three geographies, and none of them had a readme file that was current. We've been that, all of us, right?
[22:03] Uh on the ingestion layer, this ingestion layer is where we have the batch processing uh via Event Hub. In the old world, batch pipelines were written in Scala jars, uh traditional file-based, no delta semantics. A jar would just read a flat file.
[22:19] What we did is we migrated all of this into Databricks workflows, YAML defined, version controlled, CI/CD integrated pipelines. What this gives is the operational visibility. If something breaks now, we know in minutes and seconds, not in hours and days, which creates a huge impact,
[22:36] especially when you ask a question, the agent built on top of it, and it's not giving you information from 5 hours ago, which is no more relevant. Probably the passenger has already been rebooked and taken off in a different flight. Um I also wanted to focus a little bit more on the the medallion architecture that
[22:51] we have. So, this decision actually is something that helped us to have data from Synapse land into directly what is effective with silver and gold equivalent table. Uh ready to consume. Now, the problem here was like I said,
[23:09] if there is a schema change or a bad batch or anything in the raw state, it took a lot of time to even find out and recover from, right? What we did is we introduced a formal uh bronze silver gold layer medallion
[23:24] architecture. And you can rerun at any stage safely. You'll have a complete visibility. You'll have complete idea of where the data is actually getting changed if something fails. You have an operational dashboard to actually keep track of the failures in the pipeline. And that
[23:40] single architectural change eliminated an entire category of uh production incidents, right? On top box, you might see something called as MAX platform. Uh Abhishek will double click on it in a little bit. Uh what we are calling MAX platform is where the AI layer actually lives. This
[23:56] is where the exciting stuff happens, right? Uh ML workflow for tracking and model registry, Mosaic AI for foundation model. Uh we had model serving and vector search. What it really does and what we were able to accomplish through Databricks
[24:12] workflows here uh is share a cluster rather than spinning up an independent compute for every single job, which was another cost leak uh in the old architecture. Synapse dedicated pools were built 24/7 whether we are running a query or sitting aside.
[24:29] We measured approximately 40% cost reduction just by moving to auto-terminating clusters. This was massive in terms of when we calculate the total cost of uh optimization and ownership.
[24:44] Post that, what you see is on the consumption side, once we have the data processed through our ingestion layers to the entire organization of medallion architecture, it's ready to be consumed across multiple system. AI layer is one of it. It is used for a lot of AI ML models. It is used for a lot of BI
[25:00] reports. These reports are being shared across in terms of performance, baggage reconciliation, reporting on disruptions, servicing of passengers, etc. So, how did we really accelerate this journey?
[25:17] What we've done here is we we were able to leverage Tridant's uh libraries of uh Databricks accelerators across the four. So, like I said in the introduction, what makes us faster than most SIs in this space is that we have both the domain and the
[25:32] technological accelerators. I'll just break it down in this particular use case, how did we really do it? Starting from right from your inventory. So, we have a discovery accelerator, which actually shortens this time
[25:47] because of the metadata. 60% metadata-driven discovery. This was huge in terms of how quickly we were able to arrive at what is the current state and where do we even begin designing the solution? Second is your actual migration. We have uh migration frameworks we have
[26:04] done across the board. Like I said, we have more than 86 customers, 150 plus live Databricks use cases. That really gives us a template of what it will take in doing this migration at scale and with the intensity that we really designed this program for. Third thing is your validation and
[26:20] reconciliation. Huge impact in terms of how we would bring it we were able to bring in our accelerators and really reduce 30 to 50% of validation effort. This thing is already being done in different capacities across the industry and it's it's just
[26:36] positioning this and re-engineering it for the operations. That's where the team spend effort in, not really building a validation suit from scratch. And lastly, Agendify. We have our own conversational uh multi-agent interface, which blends
[26:53] very well with Genie. You can actually fine-tune it to your domain, in this case airport operations, so it's not giving you answers from generic travel and hospitality domain. Uh because of this conversational agentic piece, you were able to accelerate this and make sense of the AI model sitting on top of
[27:10] it fairly quickly. And just re-emphasizing, this if anything, this should be your takeaway, which you see at the bottom, is how we were able to do this complex modernization and enablement for AI three geographies in 3 months,
[27:26] 50% short in timelines, uh 25% cost reduction uh post-migration, agility across systems to go from actually predictive insights and actions from uh hours to seconds and minutes, and finally, have a AI-driven
[27:42] recommendation. With that being said, I wanted to hand it over to Abhishek to talk about the exciting stuff from Amadeus side. All right, so now that we have had our foundations ready uh with an with an unified governance layer, our next objective was to work towards the
[27:59] advisor, the operations advisor as we call it. We did a soft launch last year in the name of Garv, um which meant pride in uh in the Indian subcontinent languages. And um we now have rebranded the solution as Amadeus Max for aviation
[28:14] operation. Amadeus Max is a and uh it's a AI assistant and advisor across the Amadeus portfolio. And what we currently focus on is purely on aviation operations.
[28:30] So, uh Garv is our um AI-powered operations advisor, one of the flagship products. It's um it is designed to give uh the airport and airline operation uh teams the ability to query complex uh live operational data in real time in natural language or speech. Speech is
[28:46] quite important and um one of the important features Uh and quite imperative for operations. And they receive contextual and actionable answers. So for example, think of what it means in practice.
[29:02] An operations manager for instance, which airlines let's say would want to ask question like which airlines had the highest passenger processing times today at the terminal. And what's the root cause? Amadeus Smart doesn't just surface the
[29:18] number. It defines outliers, explains patterns, and supports root cause analysis. So that teams can take corrective actions. So we're moving away from a reactive to a proactive mode. One of the other evolutions of the advisor is to
[29:36] introduce smart detection. So the advisor can interact and practice and alerts to the terminal manager or the controller of the operations control center. So the way we see the advisor being used is we see an advisor in the pocket of every frontline worker on the
[29:51] terminal on the tarmac or in the control center. So now that we have the foundations, you have a government unified governance, we're clear for takeoff on our agent journey. And what's next within
[30:08] our agent tech strategy is we have the advisor now in place. We moved to production a couple of months ago. We're currently testing it out with our internal teams, our customer success manager teams. Some of the benefits we've seen is the the customer success
[30:23] managers would take two to to three weeks to generate those reports because of the wide portfolio of products that we had. That is now reduced to three minutes with the advisor. So it's quite a benefit. The next step is to work towards agent tech orchestration. And this is where we're currently looking at some killer
[30:40] use cases around disruption because that's where we have multiple systems coordinating the multiple optimizers who need to really talk which do not talk to each other today. So we'll need to build that orchestration layer on top. So that's an area that we're currently going to focus on. And then this fraud
[30:57] detection piece is currently work in progress where we plan to have an initial rollout later this year. And finally we're also working on the Arabs data exchange platform. It's a collaborative platform for airlines, airports, ground operators and their
[31:12] partners to come in, access, publish, subscribe to the to data sets available in the operational universe. One of the major challenges that we face today is that there is a lack of trust in terms of data exchange because an airline exchanging data with an airport has no visibility on how how data is
[31:30] used, for what particular use case, for how long. And we want to solve those issues from a tech tech perspective. So we have introduced workflows to manage contractual workflows, visibility on data quality, and and and provide
[31:45] end-to-end data lineage and access. So that's something that's under works as well. And of course we're going to leverage data bricks because we from a collaborative perspective we see the lot of see a lot of benefit of leveraging clean rooms as an example.
[32:01] So thank you. I think that's that's about it. So if you want to know more about our solutions, about the migration, please scan the QR code put in your names and your asks and we will get back to you. Thank you. Thank you so much guys.

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.