I am Zara. I am 19 years old. On the 18th of August 2021 I launched on a journey to fly solo around the world in a Shark ultralightaircraft. On the 20th of January 2022, 155 days later, I landed back in Wevelgem, Belgium.

With this successful circumnavigation, I am happy not only to break the Guinness World Record of youngest woman flying solo around the world, but also to reduce the gender gap by 11 years between the current youngest male record holder Travis Ludlow, 18 at the time of his record, and the previous female record holder Shaesta Waez, who was 30 when she completed her "Dreams Soar” around-the-world flight.

Not only am I hoping to reduce the gender gap in Aviation and Science, Technology, Engineering and Mathematics (STEM), but also to encourage girls and young women to pursue their dreams.

The Cannonball Run is the most notorious, illegal road race in the world. Participants race 4500km (2800 miles) on public roads from New York to Los Angeles. Arne Toman holds the overall world record at 25 hours and 39 minutes for the fastest time ever recorded in this race.

Arne will tell us the story of how he got into the underground world of Cannonballing, share details about his Cannonball cars, the strategies and hurdles in such a feat and how he broke the record not once, but twice. In addition he will show new footage from his record setting runs across the United States of America.

I am doing motorsport since I am 4 years old. From karting to touring cars, Formula 4, Formula 3, DTM - but also the Mille Miglia with a Mercedes 300 SL - at the age of 21 I have driven the fastest and most fascinating racing cars. I'm currently racing an LMP2 Protoype in the European Le Mans Series and the 24H of Le Mans, the biggest racing event in the world. But what is it that fascinates me? Is it individual sport or team sport? Is talent measurable by data analysis? What is this excitement that makes a racing driver risk his health and life at the limit in every lap? What is behind driving a racing car at an average speed of over 220 km/h at the limit? You will learn facts that will surprise you.

In a spectacular, multivisual show, Jonas talks about his great adventure, gives insights into his mentality and motivation, his objectives and how he deals with setbacks. Jonas Deichmann is the only person to have set cycling records on all major continental crossings and circled the world in triathlon. He is a multiple world record holder and best-selling author. Following his talk, Jonas will answer questions.

Over the past couple of years, Single Page Applications (SPAs) – built with React, Vue.js, or Angular – have become increasingly popular. SPAs are used from highly interactive applications to ecommerce sites and even on blogs. But they come at a cost.
There are signs that the pendulum may be swinging back again towards Multi Page Applications (MPA). We will have a look at some approaches and projects on the spectrum between classic SPA and classic MPA.

An organization's decision to move towards a central messaging platform such as Apache Kafka leads to a massive change in how one treats interfaces and data. BestSecret is an extremely fast-growing enterprise, and we as its architects face the challenge of providing technical solutions that not only support the growth of the company but also drive organizational change towards structures that are also elastic and scalable.
In this talk we will address how we tackle technical complexity in a fast-paced and ever-growing business while also helping to shape the mindset of the organization into one in which our architecture can thrive.

Making music is one of the most mysterious forms of human activity. Can computers be creative as humans are? In his talk, Matthias Röder is giving an insight into the spectacular project of finishing Beethoven's unfinished 10th Symphony with an Artificial Intelligence. He demonstrates how co-creating with machines is the new thing in music and provides an outlook in the future of making music.

Developing new drugs is currently extremely slow and expensive; but this will change soon. Modern machine learning can be used to tame the complexities of biology and dramatically improve the process. I will talk about the two complementary technologies that my company is working on. For the first, we followed the path that was laid down by the deep learning revolution in machine vision. We took a black-box approach which aims to directly relate the structure of a molecule to its pharmacological properties, and supercharged it with our own deep neural net architecture and enough combined data to enable end-to-end training. This produced a powerful new tool with high accuracy and wide applicability. The second approach that we are developing combines physics simulations with techniques borrowed from reinforcement-learning to gain a much more detailed understanding of biochemistry. This can produce accurate predictions without requiring specific experimental data for training, and could set the ground for a whole new range of machine-learning applications.

What can artificial intelligence do today? In this talk, we will use a variety of live examples to show whether the term is an over-hyped buzzword or the key technology of the digital future.

The TNG Innovation Hacking Team has been working on numerous AI projects in the field of computer vision and natural language processing for years. Thomas Endres, Martin Förtsch and Jonas Mayer will take you on a journey through the world of artificial intelligence and its architecture. Using various innovation projects, the speakers will shed light on the basic structure of the underlying neural networks, but also on their limitations. A firework of live demonstrations and showcases will round off the entertaining experience. There will be neural networks for generating art, deepfakes in real time, an NLP chatbot and an AI that can generate social media comments.

In our professional lives, we face new and complex challenges every day. AI has been one tool that helps us in this ongoing endeavour and will continue to do so. However, how we interact with AI over the decades has changed significantly and while some best practices still apply, others have lost their value. A long time has passed since the early days of Machine Learning and even since the advent of Deep Learning. In the age of Large-Scale Models, a new way of interacting with AI with new best practices lays before us. In this talk we will take a look at the new paradigms that Large Language Models present and how we can use these new paradigms to creatively solve problems.

Who wrote the Bible ? Did Shakespeare or Molière author their plays ? Who hides behind the pseudonym Elena Ferrante ? At the crossroads of scientific domains as diverse as statistics, artificial intelligence, linguistics, forensics or history, specialists of “authorship attribution” and “stylometry” are able to give more and more accurate answers to century-old historical controversies or contemporary enigmas. By taking a deep look into the mathematical properties of someone’s language, they can help identify the author of a text, no matter its nature. Wonderful poetry, ransom letters or social media posts… : all of these give a large number of clues on their authors, betrayed by their use of hardly noticeable details (function words, suffixes…) and more generally, by complex patterns in their language. We explain in this talk how this discipline grew during the past 150 years, and how our own research shed lights on enigmas from the past, from the Troubadours to Molière, or from today, as in our recent article on the authors behind QAnon’s infamous “Q” with the New York Times.

In chess, Go or heads-up poker, artificial intelligence has already triumphed over humans. However, games with more than two players and incomplete information remain a great challenge. One of these games is the Bavarian card game Schafkopf, a trick-taking game played by four players.
In his master's thesis, Alexander Sedlmayr investigated how different machine learning approaches can be applied to this game.
Franz Pfeifroth, another passionate Schafkopf player, meanwhile developed his own Schafkopf app and then also started to train an AI for this game. His question was: How does the AI work? What can it do and what can it not do? Is there a self-learning AI for a manageable game like Schafkopf? To approach these questions, he first developed a neural network for the Schafkopf bidding process. His talk deals with the problems and solutions for these. Examples of problems are: What do I have to do for the AI to "reliably" recognise that the caller of an ace must have at least one card of the suit? Can the AI distinguish between a very good solo and a solo tout?
Based on Alexander Sedlmayr's master's thesis, Franz Pfeifroth then turned to a neural network for game execution, which will also be discussed in their talk.
Finally, the current progress on the topic of "self-learning Schafkopf AI" using Monte Carlo Tree Search will be presented.

How good can an AI-generated PowerPoint presentation be? We tried it out and found out: Not bad at all.

In the age of digitalisation and AI, many tasks in our everyday lives can be automated very easily. But until recently, the biggest burden of the common business person, the PowerPoint presentation, had to be tediously created and delivered by hand. But that's over now! The TNG Innovation Hacking Team presents the PowerPoint Ghostwriter. This AI system only needs a title and automatically generates a complete slidedeck, from the headlines to the matching images. Even the presentation is handled by the artificial intelligence: A deceptively real deepfake avatar - trained with the knowledge of the entire internet - eloquently guides the listener through the slides.

In their presentation, Thomas Endres, Christoph Bergemann and Jonas Mayer give a technical overview of the PowerPoint Ghostwriter and the underlying artificial intelligence. Using illustrative live demos, they will show how GPT-3 can generate headlines and content texts in German and English, how Dall-E Mini fills the slides with appealing images and how to get a Deepfake avatar to read out slide content with the help of Amazon Polly.

Are you interested to understand, how you can modernize a monolith with several million lines of Delphi code to a SaaS solution? Learn more about our technical and organizational factors for success and how we use the strangler pattern successfully in real life. This presentation will give you detailed insights about zvoove’s road to SaaS:

- Technologies

- Deployment & Architecture

- Organisation

TNG supports BMW in designing, implementing and operating a company-wide global Cloud Data Hub on Amazon Web Services.
With petabytes of data and more than 1000 AWS accounts, there are a lot of challenges - technical as well as organizational. This talk gives an overview of how a large scale data lake can be implemented in a cloud-native way. We talk about architectural considerations, technical choices and organizational approaches.

What happens to the second largest chess website in the world when everyone is at home and suddenly wants to play chess? As a highly distributed non-profit and open source project, we were faced with both technical and organizational challenges. We needed to adapt and scale up to sustainably deliver on our mission: To promote and encourage the teaching and playing of chess all around the world - 100% funded by donations.

Speedrunning is the act of completing a video game as quickly as possible. Highly skilled speedrunners zip through levels at inhuman speeds, sometimes warping straight to the end credits. They set world records for millions of enthusiasts to watch in awe. But how does this black magic come to be? Behind the scenes, an entire community works together to discover new glitches and to further the understanding of these beloved games using a surprisingly scientific process of reverse engineering. In fact, science and video games become one and the same when trying to answer the ultimate question: what is the perfect speedrun?

After years of low inflation and the risk of drifting in a deflationary environment, inflation rates have surged in recent quarters in most regions.

In the first part of the presentation we will analyse the sources for the increase in inflation. Then we will discuss the outlook for inflation and the consequences for economic developments and financing conditions.

In that respect the focus will be on the impact of inflation expectations and the role of monetary policy in the transmission process.

The question of the origins of the current pandemic, the world's largest catastrophe since the end of World War II, is crucial for targeted measures to prevent similar or even worse pandemics in the future. What is the current state of knowledge about the origins of the pandemic? What have virology, computer science, science in general, the media, and politics contributed to resolving this question? This talk will provide a critical review of the history leading up to the Corona pandemic as well as of the events since late summer 2019 and derives necessary conclusions.

Allianz Direct, as we like to call it, a Tech company with Insurance license, embarked on Data Mesh journey in 2021.

In this session Martin Günther (Lead Data Engineer @ Allianz Direct) and Vaibhavi Rege (Platform Product Owner Data @ Allianz Direct) will present their experience, share insights and lessons learnt from their journey at Allianz Direct conceptualizing & implementing Data Mesh.

This talk looks over the design challenges around bringing NFTs to the chia's coinset model blockchain. It covers some comparisons with other blockchains, a somewhat-deep technical dive, and lessons learned from the process.

In this talk I’ll introduce a new Hypercore Protocol based app me and my team have been working on in the last year. The app is launching this spring, before this conference, but expect cool demos, and lots of P2P innovation to help tackle some of the problems we have today as Internet users.

Non-Fungible Tokens are a way to translate the idea of uniqueness and exclusivity into the digital world. Together with the idea of blockchains, NFTs can be built in a trustless manner and independent of any central supervising entity. In the year 2021, NFTs experienced a boom in the Ethereum ecosystem, mostly in the form of digital artwork and collectibles. However, there are some major challenges with the state-of-the-art, both with the underlying technology and the end user experience. In this talk, I want to highlight shortcomings of the current NFT ecosystem from a technical and an end user standpoint. In addition, I want to explain how the new NFT standard on the Chia ecosystem tackles these issues and opens up this technology for users that have previously avoided it for good reasons.

In dynamic gliding, very high speeds are achieved in glider-only flight without an engine. The planes are controlled remotely and are subjected to tremendous forces. They accelerate at an average of 60 times the force of gravity (60G) and at peak acceleration of up to 120G to previously unimagined speeds equivalent to those of an Airbus A380 with 128,000 hp! Together with his partners, Dirk Pflug has set more than 15 new world speed records since 2007, surpassing the speed of the previous year on average every year. With his latest design, speeds are targeted to exceed 1,000 km/h for the first time. The talk will provide some exciting background to this development.

Roborace and the Indy Autonomous Challenge, two autonomous racing vehicle competitions, have significantly impacted the research community around autonomous vehicles in recent years. This talk will share the story of the TUM Autonomous Motorsport research team, from running first laps at low speeds with safety drivers to going fully driverless at 270kph in a head-to-head race at CES 2022. It will discuss the overlap between the challenges faced in the racing application and in the development of robotaxis, autonomous trucking and off-road applications. In addition, it will outline how these learnings impacted the approach to autonomous driving at driveblocks and how companies can benefit from those competitions.

Attacks always get better, so your threat modeling needs to evolve. Learn what's new and important in threat modeling in 2022. Computers that are things are subject to different threats, and systems face new threats from voice cloning and computational propaganda and the growing importance of threats “at the human layer.” Take home actionable ways to ensure your security engineering is up to date.

Many of the existing security measures are either not used at all or are used incorrectly by end users. In addition to the frequent lack of awareness of the dangers on the Internet, this is often due to the lack of usability of these measures. In this talk, the Human Centered Security by Design approach will be presented in order to achieve better or appropriate usability for future security measures. Furthermore, a framework that concretely supports the providers of security measures in the creation of security interventions will be introduced and explained.

The OSS Review Toolkit (ORT) is a suite of tools that aim to assist with the automation of Open Source compliance tasks. It was founded 5 years ago as an Open Source project itself, according to the motto "Open Source compliance with Open Source tools". This talk gives an introduction to the tools in ORT, and shows how to set up a basic CI pipeline to perform compliance checks and create compliance assets.

In recent years, the development of quantum computers has been perceived primarily as a threat to IT security: Established cryptographic methods are becoming obsolete and data stored in a stockpile will be decrypted effortlessly in a few years. In contrast, this talk will focus on the positive potential of quantum computers and especially quantum machine learning (QML) for IT security and beyond. We will start with a short introduction to what QML actually is and how it works. We then look at similarities of phase transitions in physics and anomaly detection, some interesting applications and what the potential quantum advantage actually is.

Though early in its development, quantum computing is now available on both real hardware via the cloud through IBM Quantum, and also via the world’s first full commercially integrated quantum computer, IBM Quantum System One. This radically new kind of computing holds open the possibility of solving some problems that are now and perhaps always will be intractable for “classical” computers.

In this talk, we’ll discuss the history and motivation for quantum computing, how it will enhance certain types of computer problems, and the types of problems where quantum computing shows promise, such as in quantum chemistry, machine learning, and AI.

We’ll describe the basics of the technology based on IBM’s approach using superconducting transmon qubits, and discuss the IBM Quantum roadmap for reaching quantum advantage: the point where quantum computing shows demonstrable and significant advantage over classical computers and algorithms.

We will also get our hands dirty with a short demo of IBM Quantum Composer and Lab; free cloud-based tools to start coding right away, with drag and drop UI and Python-based Qiskit/OpenQASM lab environment through Jupiter Notebooks. You can dive in at the level of detail that suits your role as kernel-, -algorithm- or model developer: from basic quantum circuit and algorithm exploration to direct qubit pulse control, or why not Qiskit metal for designing your own quantum chip.

This is the story of the odyssey of two scientists whose company NanoTemper became a global leader in biophysical measurement technology. Join us on our journey to a future where we aim to create knowledge from data. A future in which our intelligent tools make it possible to treat previously untreatable diseases.
It is a story about innovations and physics, data generation and data interpretation for biotech, failures and learnings. The foremost it is a story about good and bad advices, about people, emotions and purpose.

Let me inspire you, to live your own deep tech adventure.

The first image of a supermassive black hole located in the center of our Milky Way was taken with radio telescopes and it was recently released by the Event Horizon Telescope (EHT) Collaboration. This image is another breakthrough not only for astronomers. It follows decades of observations of Active Galactic Nuclei and relativistic jet structures. I will discuss how these results were obtained by our international team of researchers and explain why these results matter.

Physicists have been working on fusion for a long time and after Lawsons theorem there was even a second law of fusion rumoured – “fusion is always 25 years away”. But recently the private sector has gotten involved with new concepts and the 25 years became 20, 15 and then 10 or even less to demonstrate net gain. However, most companies are working on the physics side and do not solve some of the major engineering obstacles to get fusion power on the grid. First Light Fusion has a new approach that sidesteps those engineering challenges and recently demonstrated the proof of fusion, independently validated by the UK Atomic Energy Authority. This opens up a path not just to net gain but also to realistic Fusion Power on the grid in the next 10-15 years.

Who doesn't dream of a city where you can get around easily at any time - without traffic jams and emissions? In urban areas, space is a limiting factor alongside noise and environmental pollution.
WIth the motivation to change that, the EDAG Group's developed the "EDAG CityBot" mobility system. The CityBot does not belong to any specific vehicle category. Rather, it represents a holistic, networked and radically different public mobility offer. EDAG CityBots are connected, autonomously driving robotic vehicles coupled with various utility modules to master all transport and work related tasks in urban areas. Constantly on the move and therefore maximally efficient. In a separated area without mixed traffic, the EDAG CityBots are coordinated by a control software and moved strategically and with foresight like pieces on a chessboard. This avoids traffic jams, reduces emissions and accidents, and reclaims public space for the expansion of cycle paths and residential or adventure space.

Supported by funding from the BMDV (project: Campus FreeCity), further prototypes of the CityBots are being developed and built in order to study the networked fleet of modular robot vehicles in a test environment in the Arena of IOT in the Deutsche Bank Park (Eintracht Frankfurt) in real operation.
The focus is on the qualification of the hardware and software of the system, in particular on the implementation of the control software for the coordination of the CityBots, the networking of the vehicles with the infrastructure, teleoperated driving and on the certification of this new vehicle class.

I make YouTube videos about math, meaning I need to compete for people's attention next to a nearly bottomless array of the world's most engaging videos, available for a viewer to click on at any moment. Of course, the need to explain math extends well beyond the usual confines of schools and online education; anytime new technology is explained, it's often the mathematically heavy parts where holding a listeners attention becomes trickiest. This talk will go through several principles I keep in mind when trying to popularize math, general enough that they should apply to most cases where you might find yourself needing to explain something technical to others, and keep them engaged all the while.

Die Zahl der Unternehmen, die sich für agile Methoden entscheiden, steigt weiter an - in vielen Unternehmen ist die agile Methode die "Standardmethode" für Teams. Weniger klar ist jedoch, warum so wenige Organisationen es geschafft haben, über die agile Bereitstellung hinaus zu agilen Organisationen zu skalieren. Seien Sie dabei, wenn wir die Einschränkungen, Herausforderungen und Ideen untersuchen, die uns daran hindern, agile Methoden von Teams zu Organisationen zu entwickeln.

Developers spend a lot of time searching code, looking for function names, classes, uses, examples. This can be done via traditional text search (grep or regular expressions) or language-aware search as part of an IDE. Code search is often the backbone of tools for automated refactoring, linting, security or static analysis.

Search becomes more challenging as repositories become larger, and as projects split across multiple languages and code generation steps. At this scale, simple text search often breaks down, becoming too imprecise or too slow.

This talk describes how the Glean open source code indexing system, developed at Meta, is used to power accurate, precise semantic code search across the largest code bases, and across multiple languages, improving developer productivity at scale.

Social media has transformed everything about our day to day lives. It provides many opportunities to extend education beyond the classroom. Tiktok provides a unique opportunity to reach a wide range of users while they are relaxed and fully engaged with their content. Educators are flocking to social media in their free time to create accounts to help as many students as they can.