Charting new heights: Keyvan Sadri from Lufthansa Group Digital Hangar on AI's role in aviation's current development and its future

Charting new heights: Keyvan Sadri from Lufthansa Group Digital Hangar on AI's role in aviation's current development and its future

Jarek Jarzębowski - July 21, 2024

The aviation industry has soared on the wings of technological advancements for decades, but recent AI breakthroughs are set to turbocharge its evolution. Airlines are now harnessing artificial intelligence to streamline flight planning and operations, reducing congestion and carbon footprint along the way. In 2023, the number of passengers carried by air in the EU almost reached the pre-pandemic levels, bringing back the need for efficient mass solutions.

But something else changed on the way - we now have broad access to powerful large language models. What does it mean for the industry? In a discussion with Jerzy Jarzębowski, Keyvan Sadri reveals the sky-high potential of AI in aviation, drawing from his hands-on experience with projects for Lufthansa. Buckle up and explore his insights into how AI is reshaping the future of flight!

Key Takeaways from the Conversation

Board up the AI journey with a solid data system: Recognizing data science needs is crucial for successful AI implementation. To implement a model, you should develop a solid data architecture to easily scale your project in the future. Think about it beforehand, especially if you need to include external data sources. Key points to consider? Your system should be fast, sustainable, and technologically up-to-date.

The sky’s the limit… or so can be your model: Whether you choose a large or small language model, there will be risks involved. Small models require a lot of work and attention to prevent limited context and legal implications. On the other hand, large language models may lack transparency and create dependencies on external providers. In the end, it all comes down to weighing the pros and cons carefully.

Data management is what you need: Maintaining a proper data warehouse will give you a solid foundation for creating and maintaining an AI solution. Data visualization and fast information extraction will help you make the best use of the data available. Once you have this resolved, you’ve prepared the ground for advanced analytics and prediction systems to step in.

Conversation with Keyvan Sadri

Jarek Jarzębowski: Let’s start with a bit of background. Can you tell us a little bit more about yourself and your expertise?

Keyvan Sadri: Sure! My name is Keyvan Sadri. I am originally a physicist and did my Ph.D. at the Max Planck Institute in Heidelberg. After finishing my Ph.D., I started working as a data science consultant in the energy industry, financial services, retail, and manufacturing. I’ve done several projects, and my last consulting job was at KPMG, where I was a consultant for four years. Then, I joined Lufthansa as a senior manager in data science, focusing on developing and delivering fraud AI and advanced analytics products related to our customer domain.

Jarek Jarzębowski: Can you briefly tell us a bit more about the product you are working on?

Keyvan Sadri: Lufthansa, which I work for at the moment, has started an ecosystem called the Digital Hangar. The aim was to develop customer-centric products, for example in customer service, the travel experience, and our B2B clients. 

I am the Product Owner of the AI team, where we develop AI products for various parts of this environment. We have products that identify personal information and documents to enhance our GDPR compliance efforts. Moreover, we have also delivered the Claim Summarizer which is our first Gen AI powered product. It can extract facts from customer claims.

Jarek Jarzębowski: Great! It’s a great variety of products with different use cases. I’m also wondering if you could give us a short explanation of your point of view on the landscape of AI in logistics and airlines in general. What is the current state of AI, and what advancements have been made in recent months or years?

Keyvan Sadri: Let me start with the challenges in the airline, especially the passenger airline industry. After the pandemic, many countries removed travel restrictions, which made the demand for flights surge. At the same time, there were many operational problems. For example, many airlines had parked their aircraft and needed to perform maintenance to ensure they were ready to fly. Additionally, many people had left the industry. As a result, the airlines had to hire new staff, including cabin crew, amidst high demand.

These operational problems put significant pressure on customer service. Customers had to rebook flights, call customer service centers, and locate lost baggage, which are all parts of the customer journey. To address these issues, Lufthansa started the Digital Hangar to improve parts of the customer journey by offering digital products. Digitalization, including AI, can help enhance customer service, rebooking processes, claim management, and even meal selection before flights.

Jarek Jarzębowski: Are you using AI and technology for demand forecasting, such as predicting meal requirements or other aspects?

Keyvan Sadri: Yes. Demand prediction is a well-known use case in network planning. Like many other airlines, Lufthansa uses machine learning to enhance the demand prediction.

Jarek Jarzębowski: Can you share some challenges you have encountered while working on these AI projects?

Keyvan Sadri: One major challenge is GDPR compliance, but we handle it positively, ensuring clear guidelines on what we can and cannot do with data. Another challenge is the scalability of AI solutions. When a model goes into production, it must consistently deliver results, whether for decision support or automation. Many companies develop models that never reach production, leading to wasted investment. Lufthansa is well-positioned to produce AI models, but it still requires time, money, and standardized processes.

Jarek Jarzębowski: What is the biggest challenge in putting a model into production?

Keyvan Sadri: The main challenge is companies often do not recognize data science needs its own environment and ecosystem. Data science involves understanding the business need, finding the right data, having proper data engineering and architecture, and establishing a sustainable MLOps process. Many initiatives start with hiring data scientists without considering the broader ecosystem needed for success. Companies must view data as an ecosystem to realize value from AI.

Jarek Jarzębowski: What best practices should companies follow for data management and AI implementation?

Keyvan Sadri: Based on my consulting experience, using data involves three steps: data management, business intelligence, and advanced analytics and machine learning. First, companies should manage data effectively, collecting it from all sources and maintaining a proper data warehouse. Next, they should use business intelligence to understand their data through visualization and fast information extraction. Finally, they can move to advanced analytics and machine learning for prediction and decision support.

Jarek Jarzębowski: Should companies know their end goal when starting with data management, or can they apply best practices and use the data later?

Keyvan Sadri: In the first step, companies should at least know their own data and collect it correctly, ensuring a sustainable system that is technologically up-to-date and fast. While they might not need specific data science use cases initially, having a good data architecture will support future data science projects. External data sources will also be necessary for advanced analytics, and a well-designed data ecosystem can handle this.

Jarek Jarzębowski: What do you think about the future of AI in aviation and logistics? How should companies prepare for it?

Keyvan Sadri: The future holds many possibilities, such as updating prediction methods with more data and advanced techniques, using the Internet of Things for maintenance, and leveraging generative AI for text and voice processing. Companies should be careful with the cost and implementation of generative AI, ensuring defined business value. Smaller language models can be useful for specific tasks, reducing dependency on expensive external providers. Preparing involves focusing on defined use cases and exploring small language models.

Jarek Jarzębowski: Are there any risks with using large language models versus smaller, internal ones?

Keyvan Sadri: Large language models can be expensive and create dependencies on external providers. Training small language models is a significant investment, requiring proper deployment and monitoring. Risks include hallucination or wrong information, which can have legal implications. Large language models lack transparency, making it difficult to understand their decision-making process. Companies should be cautious about these risks and ensure robust monitoring and validation.

Jarek Jarzębowski: Apart from data management and MLOps, what else should companies focus on when introducing data science and AI?

Keyvan Sadri: Companies should balance having strong business cases with enabling innovation. Over-focusing on business value can stifle innovation. Companies should find ways to encourage innovation, such as dedicating time for experimentation and innovation sprints. In the Digital Hangar, we use a SAFE framework with quarterly innovation sprints to foster creativity and new ideas.

Jarek Jarzębowski: Can you share some successes from innovation sprints or hackathons?

Keyvan Sadri: We’ve seen significant success from innovation sprints and hackathons. These events bring together people from different domains within Lufthansa to work on new ideas. While I can’t share specific product details, I can say that these initiatives have led to valuable insights and the development of impactful products.

Jarek Jarzębowski: Thank you for sharing your insights and experiences, Keyvan. This has been a very enlightening conversation.

Keyvan Sadri: Thank you, Jarek. It was a pleasure discussing these topics with you.

Keyvan Sadri’s Background

Keyvan Sadri is an expert in AI and data science with a Ph.D. in Theoretical Chemistry from Universität Heidelberg and a Master’s in Physical Chemistry from Sharif University of Technology. His career spans multiple industries, including roles as a Senior Data Scientist at KPMG Deutschland and a consultant for BearingPoint and Publicis Sapient, where he led various big data and AI projects.

Currently, Keyvan collaborates with Lufthansa, serving as the AI Product Owner and Interim Lead of the Community of Practice for Data Science and Machine Learning. His work focuses on operationalizing AI models, optimizing flight operations, and enhancing customer service through AI innovations, ensuring the aviation industry remains at the forefront of technological advancements.

About Lufthansa Group Digital Hangar

The Lufthansa Group Digital Hangar aims to make flying easier and more enjoyable by combining physical, human, and digital experiences across all Lufthansa Group airlines. The project focuses on digitalizing every part of your journey, from booking and check-in to in-flight services and baggage handling. Working on creating better online shopping and booking experiences, they drive personalization efforts to ensure you get the right information and offers at the right time. Their goal? Leading the way in digital solutions for the airline industry.

Lufthansa Group Digital Hangar

Lufthansa Group Digital Hangar

Closing Thoughts

The aviation industry is regaining strength after pandemic-induced setbacks. However, this means not only business opportunities but also challenges. Especially in the era of climate change, airlines must focus on data-driven solutions that continuously optimize their operations in terms of fuel consumption, waste generation, and overall efficiency. 

At the same time, passenger expectations are rising, and airports must handle an increasing number of airlines and connections. This, in turn, leads to delays, impacting customer satisfaction. A data-driven approach involving advanced AI predictions can help airlines manage fluctuating demand and meet these expectations. 

Will the aviation industry reach new heights with expanding AI capabilities? Everything suggests that the coming years could be game-changing. Let’s wait and see what Lufthansa has been working on!

About the author

Jarek Jarzębowski

Jarek Jarzębowski

People & Culture Lead

Linkedin profile Twitter

Jarek is an experienced People & Culture professional and tech enthusiast. He is a speaker at HR and tech conferences and Podcaster, who shares a lot on LinkedIn. He loves working on the crossroads of humans, technology, and business, bringing the best of all worlds and combining them in a novel way.
At nexocode, he is responsible for leading People & Culture initiatives.

Would you like to discuss AI opportunities in your business?

Let us know and Dorota will arrange a call with our experts.

Dorota Owczarek
Dorota Owczarek
AI Product Lead

Thanks for the message!

We'll do our best to get back to you
as soon as possible.

This article is a part of

AI Revolution Diaries
6 articles

AI Revolution Diaries

Step into the narrative of change with our AI Revolution Diaries, where each interview captures a moment in the ongoing revolution of artificial intelligence across industries. These diaries detail the firsthand experiences of businesses at the forefront of integrating AI, highlighting the transformative impact and the lessons learned throughout their journey of innovation.

Engage with our series to discover the strategies that drive successful AI integration, and grasp the benefits and hurdles encountered by pioneers in the field. Let us be your guide in navigating the transformative journey of AI, empowering your business to harness the full potential of data and shape the future of your industry.

check it out

Becoming AI Driven

Insights on practical AI applications just one click away

Sign up for our newsletter and don't miss out on the latest insights, trends and innovations from this sector.

Done!

Thanks for joining the newsletter

Check your inbox for the confirmation email & enjoy the read!

This site uses cookies for analytical purposes.

Accept Privacy Policy

In the interests of your safety and to implement the principle of lawful, reliable and transparent processing of your personal data when using our services, we developed this document called the Privacy Policy. This document regulates the processing and protection of Users’ personal data in connection with their use of the Website and has been prepared by Nexocode.

To ensure the protection of Users' personal data, Nexocode applies appropriate organizational and technical solutions to prevent privacy breaches. Nexocode implements measures to ensure security at the level which ensures compliance with applicable Polish and European laws such as:

  1. Regulation (EU) 2016/679 of the European Parliament and of the Council of 27 April 2016 on the protection of natural persons with regard to the processing of personal data and on the free movement of such data, and repealing Directive 95/46/EC (General Data Protection Regulation) (published in the Official Journal of the European Union L 119, p 1); Act of 10 May 2018 on personal data protection (published in the Journal of Laws of 2018, item 1000);
  2. Act of 18 July 2002 on providing services by electronic means;
  3. Telecommunications Law of 16 July 2004.

The Website is secured by the SSL protocol, which provides secure data transmission on the Internet.

1. Definitions

  1. User – a person that uses the Website, i.e. a natural person with full legal capacity, a legal person, or an organizational unit which is not a legal person to which specific provisions grant legal capacity.
  2. Nexocode – NEXOCODE sp. z o.o. with its registered office in Kraków, ul. Wadowicka 7, 30-347 Kraków, entered into the Register of Entrepreneurs of the National Court Register kept by the District Court for Kraków-Śródmieście in Kraków, 11th Commercial Department of the National Court Register, under the KRS number: 0000686992, NIP: 6762533324.
  3. Website – website run by Nexocode, at the URL: nexocode.com whose content is available to authorized persons.
  4. Cookies – small files saved by the server on the User's computer, which the server can read when when the website is accessed from the computer.
  5. SSL protocol – a special standard for transmitting data on the Internet which unlike ordinary methods of data transmission encrypts data transmission.
  6. System log – the information that the User's computer transmits to the server which may contain various data (e.g. the user’s IP number), allowing to determine the approximate location where the connection came from.
  7. IP address – individual number which is usually assigned to every computer connected to the Internet. The IP number can be permanently associated with the computer (static) or assigned to a given connection (dynamic).
  8. GDPR – Regulation 2016/679 of the European Parliament and of the Council of 27 April 2016 on the protection of individuals regarding the processing of personal data and onthe free transmission of such data, repealing Directive 95/46 / EC (General Data Protection Regulation).
  9. Personal data – information about an identified or identifiable natural person ("data subject"). An identifiable natural person is a person who can be directly or indirectly identified, in particular on the basis of identifiers such as name, identification number, location data, online identifiers or one or more specific factors determining the physical, physiological, genetic, mental, economic, cultural or social identity of a natural person.
  10. Processing – any operations performed on personal data, such as collecting, recording, storing, developing, modifying, sharing, and deleting, especially when performed in IT systems.

2. Cookies

The Website is secured by the SSL protocol, which provides secure data transmission on the Internet. The Website, in accordance with art. 173 of the Telecommunications Act of 16 July 2004 of the Republic of Poland, uses Cookies, i.e. data, in particular text files, stored on the User's end device.
Cookies are used to:

  1. improve user experience and facilitate navigation on the site;
  2. help to identify returning Users who access the website using the device on which Cookies were saved;
  3. creating statistics which help to understand how the Users use websites, which allows to improve their structure and content;
  4. adjusting the content of the Website pages to specific User’s preferences and optimizing the websites website experience to the each User's individual needs.

Cookies usually contain the name of the website from which they originate, their storage time on the end device and a unique number. On our Website, we use the following types of Cookies:

  • "Session" – cookie files stored on the User's end device until the Uses logs out, leaves the website or turns off the web browser;
  • "Persistent" – cookie files stored on the User's end device for the time specified in the Cookie file parameters or until they are deleted by the User;
  • "Performance" – cookies used specifically for gathering data on how visitors use a website to measure the performance of a website;
  • "Strictly necessary" – essential for browsing the website and using its features, such as accessing secure areas of the site;
  • "Functional" – cookies enabling remembering the settings selected by the User and personalizing the User interface;
  • "First-party" – cookies stored by the Website;
  • "Third-party" – cookies derived from a website other than the Website;
  • "Facebook cookies" – You should read Facebook cookies policy: www.facebook.com
  • "Other Google cookies" – Refer to Google cookie policy: google.com

3. How System Logs work on the Website

User's activity on the Website, including the User’s Personal Data, is recorded in System Logs. The information collected in the Logs is processed primarily for purposes related to the provision of services, i.e. for the purposes of:

  • analytics – to improve the quality of services provided by us as part of the Website and adapt its functionalities to the needs of the Users. The legal basis for processing in this case is the legitimate interest of Nexocode consisting in analyzing Users' activities and their preferences;
  • fraud detection, identification and countering threats to stability and correct operation of the Website.

4. Cookie mechanism on the Website

Our site uses basic cookies that facilitate the use of its resources. Cookies contain useful information and are stored on the User's computer – our server can read them when connecting to this computer again. Most web browsers allow cookies to be stored on the User's end device by default. Each User can change their Cookie settings in the web browser settings menu: Google ChromeOpen the menu (click the three-dot icon in the upper right corner), Settings > Advanced. In the "Privacy and security" section, click the Content Settings button. In the "Cookies and site date" section you can change the following Cookie settings:

  • Deleting cookies,
  • Blocking cookies by default,
  • Default permission for cookies,
  • Saving Cookies and website data by default and clearing them when the browser is closed,
  • Specifying exceptions for Cookies for specific websites or domains

Internet Explorer 6.0 and 7.0
From the browser menu (upper right corner): Tools > Internet Options > Privacy, click the Sites button. Use the slider to set the desired level, confirm the change with the OK button.

Mozilla Firefox
browser menu: Tools > Options > Privacy and security. Activate the “Custom” field. From there, you can check a relevant field to decide whether or not to accept cookies.

Opera
Open the browser’s settings menu: Go to the Advanced section > Site Settings > Cookies and site data. From there, adjust the setting: Allow sites to save and read cookie data

Safari
In the Safari drop-down menu, select Preferences and click the Security icon.From there, select the desired security level in the "Accept cookies" area.

Disabling Cookies in your browser does not deprive you of access to the resources of the Website. Web browsers, by default, allow storing Cookies on the User's end device. Website Users can freely adjust cookie settings. The web browser allows you to delete cookies. It is also possible to automatically block cookies. Detailed information on this subject is provided in the help or documentation of the specific web browser used by the User. The User can decide not to receive Cookies by changing browser settings. However, disabling Cookies necessary for authentication, security or remembering User preferences may impact user experience, or even make the Website unusable.

5. Additional information

External links may be placed on the Website enabling Users to directly reach other website. Also, while using the Website, cookies may also be placed on the User’s device from other entities, in particular from third parties such as Google, in order to enable the use the functionalities of the Website integrated with these third parties. Each of such providers sets out the rules for the use of cookies in their privacy policy, so for security reasons we recommend that you read the privacy policy document before using these pages. We reserve the right to change this privacy policy at any time by publishing an updated version on our Website. After making the change, the privacy policy will be published on the page with a new date. For more information on the conditions of providing services, in particular the rules of using the Website, contracting, as well as the conditions of accessing content and using the Website, please refer to the the Website’s Terms and Conditions.

Nexocode Team