Infrastructure as Code? - Sure! What else?! But after we establish a common understanding for this very powerful buzzword: How do I actually apply that in my context? The variety of offered tools is huge and keeps growing. Learning curves are often steep and the trade offs between generalization and customization not obvious. With our talk, we would like to give a brief introduction on this topic. We will work out common aspects of Infrastructure as Code that are worth known when making a design decision. An exemplary selection of tools is demonstrated and compared. The goal of the talk is to convey a fundamental understanding and overview of this topic. This aims at both the DevOps engineer or a Product Owner, or whoever may face the challenge of provisioning new infrastructure.

Software and AI Startups are not your standard SW technology suppliers. This is a relation that requires a specific working relationship. We explain how Munich TUM Venture lab fosters such relations and have a practical example with leading AI startups DQC.AI.

Are you ready to take your prompt engineering skills to the next level?

In this presentation, you will dive into the fascinating world of ChatGPT. The speakers will cover the basics of GPT and the unique features of one of today's most powerful text-based dialogue systems.

Another important aspect is to increase the understanding of prompt engineering. Optimising prompts can dramatically improve the quality of ChatGPT's responses. We will cover some best practices for prompt engineering, such as chain-of-thought or declarative prompting.

In the talk, you will gain a deeper understanding of ChatGPT and get some practical tips on how to improve prompt engineering.

Dark energy - the accelerating expansion of the Universe - is one of the key unsolved mysteries of physics today. A prominent way to understand cosmic expansion and growth is through vast surveys of the night sky where hundreds of millions of galaxies are photographed and catalogued, searching for subtle correlations in their positions and shapes. At LMU and the Max Planck for extraterrestrial Physics we develop powerful machine learning and Bayesian inference pipelines to extract insight from raw data. But calibrating and validating these tools cannot be done with real observations alone! Instead we build fair and statistically valid generative synthetic data models to characterise and forecast performance for yet to be seen astrophysical measurements. In my talk I will outline the key machine learning methods which power our work within the Dark Energy Survey (DES) and the Legacy Survey of Space and Time (LSST) and highlight their real world applications which could bring business value.

The open source release of Stable Diffusion has recently caused a "cambrian explosion of creative AI tools''. In this talk, I will discuss the underlying generative modeling paradigm and the training process of Stable Diffusion. In particular, I will discuss two-stage approaches to generative modeling, focusing on autoregressive transformers and diffusion models trained in the latent space of VQGAN and related autoencoding models. I will also discuss how both approaches can be combined with flexible conditioning mechanisms using transformer architectures, resulting in powerful multimodal generative models.

Large Language Models (LLM) lead to a paradigm shift in natural language processing, chatGPT and Bard being at the forefront of that wave. In this talk, we outline how LLMs impact the insurance industry, what issues and flaws we face and how we are currently dealing with their implementation.

Augmented Reality (AR) embeds spatially-registered computer graphics into a user’s view of the real world. During the last 23 years, AR has progressed enormously from a niche technology to a widely investigated one.

This keynote consists of two parts. First, I speak about how major challenges for AR have been solved over the last 23 years. Second, I speculate about what the next 23 years are going to bring.

The goal of this keynote is to present a Birdseye view of the AR domain, including the balance of power between the major AR forces US and China. In my view, Europe has a very big, possibly almost impossible, challenge ahead to catch up. I hope that my talk will contribute to laying the seeds of a major European AR initiative.

Taking the perfect profile picture without a photo shoot or using Photoshop? It’s now possible thanks to artificial intelligence (AI).

Creating high-quality and consistent profile pictures for hundreds of employees can be a daunting task. However, artificial intelligence offers an innovative solution to create individually appealing portraits. In this talk, we present the "Portrait Editor". A software that creates professional portraits even from poor snapshots.

In recent years, AI has gained importance in many areas. In our talk, we highlight three key technologies that our Portrait Editor uses to achieve impressive results. Learn how face recognition algorithms calculate the ideal image section and how segmentation techniques efficiently remove the background. We explain text-to-image techniques to fill in missing image areas or replace them with new content.

The speakers will address the specific challenges of using these models, such as effective prompting. The audience will have the opportunity to experience the Portrait Editor for themselves in numerous live demos.

Mass spectrometry enables the identification and quantification of chemical compounds in samples. Peptides are fragmented in the mass spectrometer and the masses of the fragments are measured and recorded. Prediction of chemical and physical properties of peptides and peptide fragmentation patterns with the help of AI-powered algorithms is delivering much better results than with classical algorithms. As a result, we can identify and quantify, with high confidence, more proteins in complex samples. However, the amount of data which needs to be processed, and the prediction of peptides in complex samples requires significant computational power.

Thermo Fisher Scientific customers can benefit from the advancement of AI-driven analysis by using a highly scalable cloud service to perform these complex computations. The peptide search engine CHIMERYS (developed by MSAID GmbH) is used at the core of the service. TNG has designed the cloud service in a way that also enables on-premise deployment for customers who prefer to process sensitive data on their own hardware.

In the talk, we will briefly explain proteomics analysis and how a sample from a mass spectrometer leads to protein identification. Afterwards we want to follow up with learnings and our takeaways from a journey of implementing the CHIMERYS Platform.

When the German inventor, entrepreneur and builder of the first functioning, programmable computer, Konrad Zuse, worked on the first-ever chess program during the war years of 1942 to 1945, several advantages of such a program may have interested him: replacing a chess opponent, finding the best chess moves through precise calculations, and evaluating the resulting positions. In the early days of computers, chess programs were also interesting because they offered the opportunity to compare the capabilities of computers with those of humans. While initially, chess engines were inferior to strong human players, the 1997 victory of the chess machine "Deep Blue" against world champion Garry Kasparov marked a turning point. Since then, chess engines could only be used as opponents with reduced strength, as their playing strength far exceeds that of any human. This, however, made the engine increasingly interesting as a training tool for chess players, especially for professionals.

Driven by ever more powerful processors, engines calculate a large number of possible moves in a given position in the shortest possible time and come to almost objective evaluations in seconds, while humans would have to invest days or weeks to come to the same conclusions. They consider every move almost bias-free, leaving out traditional insights in tactics, strategy and opening theory. Consequently, the style of many players has changed in recent years. Like many inventions, the engine can also be misused. In recent chess history, there have been cases of fraud where chess engines always played the decisive role.

Current engines achieve playing strengths of around 3,500 ELO. For comparison: The playing strength of world champion Magnus Carlsen is around 2,850 ELO. In this talk, the fascination of this enormous, "inhuman" playing strength of chess engines is explored and illustrated through examples.

Valve's Half-Life is one of the most well known video game franchises in the world. Its second installment, originally released in 2004, sees players take on the role of crowbar-wearing physicist Gordon Freeman to combat an Alien invasion. The game was well received by players and critics alike, praising its diverse environments and gameplay and its clever and innovative use of physics. To this day, it remains one of the highest rated games of all time.

Almost 18 years later, our fan-made modification for Half-Life 2 [1] allows you to experience the entire game in Virtual Reality. Wearing a VR headset and wielding motion controllers, players can step into Gordon's shoes like never before, actually seeing the world from his perspective and physically swinging the crowbar around. For the past 2 years, I have been the project's lead developer. I will take you on a brief journey of the project's decade-long history and explain what it takes to make a classic game work in Virtual Reality. I will reflect on the challenges of developing for Virtual Reality - and why you should give this new way of experiencing Gordon Freeman's story a try.

[1] https://store.steampowered.com/app/658920/HalfLife_2_VR_Mod/

What are cognitive robotics?

Three important characteristics define a cognitive robot. First, the ability to fully perceive the environment - sight, hearing and even touch. Second, to be able to process these stimuli and react independently even to new and unknown things. And finally, to be able to "remember" what it has learned for the future in order to further develop existing abilities. Such robots are capable of reliably recognising humans in their vicinity. This effers completely new perspectives for our public life, for example regarding dangerous jobs or shortages of skilled workers. I myself am even convinced that we will soon have the first household robots.

What distinguishes x from other start-ups in robotics?

For one thing, we are willing to reinvent the wheel. That's the only way we managed to create not only the fastest and most precise cobot in the world within three years, but also all components such as sensors and actuators. This way we have progress in our hands and don't have to hope for innovations from suppliers. Secondly, as far as I know, we are the first company in the world to develop cognitive robots. Thirdly, we don't just make products. Rather, we offer a technology platform that partners from all over the world can plug into. Our components and robots combined with the countless ideas and know-how from a wide range of industries - this makes many special applications possible in a very short time. We call this idea the Neuroverse.

In this talk, we will explore the challenges associated with flexibility in traditional packaging methods for industrial products. The diverse range of shapes, sizes, weights, and packaging configurations of these products poses a significant hurdle for efficient packaging. However, at Robotcloud, we are developing a solution that combines advanced robotics, artificial intelligence, and machine vision capabilities to address this complexity.

Our solution focuses on creating adaptable robotic systems that can effectively handle the diverse range of products. We understand that each product may require specific gripping mechanisms, force sensing, or delicate handling to prevent damage during packaging. By integrating these capabilities into our robotic systems, we ensure that each product is handled with precision and care.

One of the key advantages of our solution is its seamless integration into existing production lines. We have designed our system to require minimal effort from the operator during the initial setup process. This allows for quick and efficient implementation, minimizing disruptions to the production workflow.

Through the integration of advanced robotics, artificial intelligence, and machine vision, our solution enables dynamic and flexible packaging processes. We have leveraged cutting-edge technologies to ensure that our robotic systems can adapt to the changing requirements of different products, providing the flexibility needed to meet diverse packaging configurations.

The vision of autonomous vehicles in the passenger car sector has existed for a long time. Many companies worldwide are developing systems for the full automation of vehicles and are making significant progress. Driver assistance systems are widespread and are developed constantly. However, despite this progress, vehicles with Level 4 or 5 automation are not yet widely used on public roads. Teleoperation, i.e. the remote control of vehicles with the aid of video data, can help to fill these crucial gaps.
In the logistics sector in particular, there is a strong demand for automated systems due to the shortage of skilled workers. In the blue collar sector, solutions are urgently needed due to demographic change and targeted growth. Driverless transport systems (AGVs) have already been available here for some time. However, the market share of autonomous vehicles in the field of industrial trucks and forklifts is still below 5 per cent. The high complexity of the technology as well as established customer processes pose a challenge to the spread of these systems. Through a system for the teleoperation of vehicles, it is already possible to implement the functionality of autonomous vehicles in this area. Here, a human driver provides support and steps in when needed.

Code indexing is the technology underlying many of the essential tools in the developer's workflow, from IDEs to code search to analysis tools. But code indexing is fragmented: it's usually done differently for each language and even each tool. Technologies like LSIF aim to unify across languages, but are aimed at specific tools or use cases. Glean is an open-source code indexing system we're building at Meta that fills the gaps: scaling to multiple languages, to very large codebases, and crucially including a flexible and powerful query language ("Angle") designed specifically for querying data about code. In this talk I'm going to cover the design of Glean and in particular its interesting query language, and also cover some of the ways that functional programming was useful in building Glean.

In this presentation, we will focus on the use of Haskell in addressing challenges related to high-speed data transmission via laser signals between ground stations and satellites. At the onset, we will discuss how atmospheric conditions induce deformations in the laser signal, which could impede data transmission. We will explain how adaptive optics, utilizing segmented deformable mirrors, have been used to counter these deformations. In addition, we will highlight the crucial role of wavefront sensors that operate at a minimum sample rate of 5 kilohertz to manage terabit-level throughput.

A substantial part of our discussion will revolve around the calibration of the mirror segments. This process involves computing 256 two-dimensional gravity points from each image within a strict timeframe of 3 microseconds. We found that this task was beyond the capability of a standard x86 processor.

To address this, we used an FPGA which was programmed using Haskell. We will describe our use of the Clash compiler to transform Haskell to low-level VHDL in order to create a suitable FPGA configuration. A key section of the talk will be dedicated to our methodical approach in Haskell, starting from a high-level specification and refining it to a detailed specification that meets the performance requirements of 3 microseconds. Our goal is to provide a deep understanding of how Haskell, in combination with the Clash compiler, can be efficiently applied in high-performance computing scenarios.

Learn how to combine the best of both TypeScript and Haskell for frontend development. TypeScript developers will learn how to take the reactive principles that made React successful to the next level with pure Functional Reactive Programming (FRP). Haskell developers on the other hand will learn how to easily integrate with the broader JavaScript ecosystem.

Links:

React: https://hackage.haskell.org/package/react
reactive-banana: https://hackage.haskell.org/package/reactive-banana,
https://hackage.haskell.org/package/reactive-banana/docs/Reactive-Banana-Combinators.html

Three decades ago, scientists began to propose quantum computing as a tool to harness the laws of quantum mechanics to solve complex problems. Today, quantum computing is a rapidly developing technology with a potential impact on many applications. In this talk, we will provide our perspective on some of these applications, particularly those related to climate and sustainability. We will also go through some of the technological challenges that quantum engineers will have to overcome in the future to bring these advances into reality.

In this talk, I will present the concept of the quantum watch that was developed during my postdoctoral project at Uppsala University (Sweden) in the HELIOS laser laboratory. The watch comprises a helium atom that is excited into a wave packet of Rydberg states using an ultrashort laser pulse with a broad energy bandwidth. Rydberg states are excited states in atoms and molecules that have long lifetimes and orbitals that extend micrometers away from the atomic core. By exciting multiple Rydberg states at the same time, it is possible to perform quantum beat spectroscopy where the energy difference between the states causes constructive and destructive interference of the photoelectron yield. When exciting multiple Rydberg states close to the He ionization potential, we discovered a remarkably rich beat structure in the time-resolved photoelectron yield (see Fig. 1 bottom). Additionally, we observed almost perfect agreement between simulations and the intricate experimental signal. Using this quantum watch, we can measure time differently, not by counting the clock's ticks as usual, but by obtaining time fingerprints, allowing for very accurate timekeeping on the femtosecond timescale without the need for a counter. The quantum watch has the potential to become a valuable tool in time-resolved spectroscopy. The study detailing our findings has recently been published in the Physical Review Research journal.

Figure Description

Simulated photoelectron yield as a function of the delay between two laser pulses.
Top: wave packet consists of two Rydberg states. No time fingerprint could be obtained.
Bottom: wave packet consists of 40 Rydberg states. Time fingerprint is created

Every year, natural disasters worldwide destroy values in the multi-digit billions. The most recent damage balance for the year 2022 amounted to approximately USD 270 billion. Munich Re's global damage data from severe weather events collected since the 1970s indicate that climate change is driving loss trends in some regions. Many of our clients are already directly affected by these risk changes, and they also expect the insurance industry to respond to the resulting challenges.

Climate risks: Understand - measure - manage

Understanding the causes and consequences of climate change, be it natural or man-made, is part of Munich Re's core expertise. We offer our clients a comprehensive and long-term strategic approach to climate risks: "Understand - Measure - Manage". Our service chain results in solutions for the individual management of (change) risks.

The ongoing development of damage simulation models is an essential element to enable an up-to-date assessment (quantification) of risks, regarding either their absolute damage potential or the risk change through climate scenarios compared to the current situation. Munich Re has expanded its range of services in recent years and now also provides this expertise via software systems ("software as a service").

Data analysis is currently heavily involved with artificial intelligence methods to identify structures. This talk shows an alternative way to interpret the data and make it accessible for analysis. It will be shown how the geometric understanding of a problem can lead to suitable methods and ultimately to new insights.

The future of medical therapies will be defined by individual, customized, and data-driven approaches. By using state-of-the-art technologies and accessing a wide variety of data sources, diseases can be detected in ways previously unimaginable. That way, appropriate therapies can be developed: A tremendous improvement for patients and significant relief for the healthcare system. The presentation provides an insight into the status and future of digitalization in the healthcare sector.

Germany is facing an energy trilemma: supply is not permanently secured, there is too little sustainably produced energy and it is too expensive. Fusion technology can serve as a complementary technology to renewable energies and therefore solve the energy trilemma. After the historic breakthrough at the National Ignition Facility (NIF) in December 2022 in the USA, a new era in fusion research has begun: the experiment has shown that laser-based fusion works - and therefore has realistic potential to produce clean electricity that is also capable of covering base-loads by the 2030s. Europe should not miss this opportunity.

Marvel Fusion, founded in Munich in 2019, is pursuing a new and fast approach to the commercialization of laser-based fusion energy. The new technology concept is based on high-intensity short-pulse lasers and proprietary fuel targets. Scientists from the Ludwig-Maximilians-University (LMU) in Munich, the University of Stanford, and the Massachusetts Institute of Technology (MIT) are working with Marvel Fusion to realize its fusion concept. Marvel Fusion aims to commercialize the technology within the next 10 years.

The human sense of touch is fundamentally important in our everyday lives, yet in media technology, it has been neglected for a long time. Today, we have sophisticated audio and video applications that have enabled immersive VR experiences. However, in VR the level of immersion is limited due to the absence of haptic cues.

In this talk, I want to present to you the new and exciting field of human haptic communication over the internet. We will look into what haptics are, where research currently stands, what I have developed and what comes next in this field.

Starting from scratch, a group of students at the TU Munich set out one goal for themselves. To build the most energy efficient car possible. With 80 team members from diverse faculties and origins, the TUfast Eco Team first developed and then manufactured a concept car designed to reach record-breaking efficiency. Two factors are especially important for this. The aerodynamics of the car and the weight.  In this talk, the TUfast Eco Team will present to you how they optimized these two variables to max out the efficiency of the car and push the boundaries of what’s currently imaginable.

Expect to learn about a team of student passionate about building an innovative car of the future and a deep dive into the creation of this car and the techniques used in the process like adjoint optimization simulation and topology optimization.

The prototype will of course also be displayed. We look forward to having you!

Lunar conditions pose challenges for handling and transporting regolith, for which current technologies fall short. To address this issue, a modular regolith transport solution comprising vertical and horizontal sections is presented. The horizontal system features an enclosed conveyor belt that effectively transports regolith up to 5 m, followed by a container with a rotating sieve mesh that filters out rocks larger than 3.5 cm. The vertical system is a tube with a fixed screw and scoops attached at the lower end that transports the regolith upwards for 2.5 meters. The rotation of the tube generates friction between the regolith and tube walls, that together with the inner screw propels the regolith upwards while the scoops push the regolith into the tube. The system benefits greatly from the lunar gravity conditions, and its confinement mitigates dust generation. An IoT-based system allows for intelligent telemetry and control to monitor dust generation, mass flow, and inflow of larger rocks. The system aims to transport one ton of material per day on Earth, with a power requirement of about 800 W and an approximated weight of 225 kg. The full-scale model of the system was designed and fabricated by the Spaceship EAC team to participate in the "Over the Dusty Moon Challenge" hosted by the Colorado School of Mines. This innovative solution effectively addresses the key challenges of transporting large amounts of regolith under lunar conditions, while minimizing dust generation and discarding large rocks.

The extreme energy requirements associated with fast round trip missions within the solar system or interstellar flight makes extraction from energy sources in space desirable. Electromagnetic interaction with the ionized gas flow in space (such as the solar wind or interstellar medium) offers the potential for significantly larger interaction areas than the physical size of the system itself. Various concepts have been developed to harness the solar wind for propulsion, including the MagSail, e-sail, and plasma magnet. These concepts primarily function as drag devices, limiting their velocities to that of the solar wind speed (~700 km/s), generating a propulsive force predominantly in the direction of the wind. In the proposed method a propellantless propulsion system, termed plasma surfing, generates a lifting force perpendicular to the apparent direction of the wind. The lift-generating mechanism extracts power from the flow over the vehicle in the apparent wind direction, which is then used to accelerate the surrounding medium in the transverse direction, generating lift at the expense of drag. Inspired by the dynamic soaring maneuvers performed by seabirds and gliders, which exploit differences in wind speed to gain velocity, the proposed technique involves the plasma surfer traveling between regions of the heliosphere that have varying wind speeds. Through this process, the system gains energy without the need for propellant and with only moderate onboard power requirements. This technique enables rapid transit within the solar system and voyages into interstellar space.

At Roche, we apply artificial intelligence (AI) and machine learning (ML) to dramatically accelerate and transform our research and development of new therapeutics, diagnostics and treatments to help patients live longer and better lives. In this talk, we will highlight the role of AI when it comes to personal healthcare, and show some examples of AI in the Pharma R&D value chain.

The Cheops Pyramid in Egypt is one of the most famous and oldest structures of mankind. It is not only the central object of Egyptian national identity, but also a World Heritage Site and one of the Seven Wonders of the Ancient World. It is the only one (and, by the way, the oldest) of these miraculous buildings that still exists. The talk will deal with the fascination with this mythical building and its exploration. The following questions will be answered: Why is the Cheops Pyramid still so interesting today? What do we learn about the sustainability of buildings? Which areas of the Cheops pyramid have been researched well? How do you examine the best-studied building in the world? What does particle physics have to do with it and what do AI and augmented reality do? What do we learn from these investigations for future research projects? And will there be further discoveries?

This talk will explore the journey of creating a robotic quadruped inspired by Boston Dynamics Spot, from the initial planning and sourcing of components, 3D printing of parts, assembly and calibration, to the training using reinforcement learning. Attendees will learn from the challenges encountered, solutions adopted, and key lessons learnt.

In this enlightening talk, Philipp Dettmer, renowned for his impactful and engaging educational content, delves into the art and science of effective science communication. His presentation aims to unravel the methods behind creating content that resonates with billions around the globe.

Dettmer elucidates how to transform complex scientific concepts into easily understandable and engaging narratives, demonstrating the power of visuals, storytelling, and simplicity. His talk further explores how to leverage these techniques to not just attract, but also retain audience attention and stimulate curiosity. Drawing from his vast experience as the founder of Kurzgesagt – In a Nutshell, a channel that has successfully bridged the gap between complex science and the general public, Dettmer shares valuable insights on the creation process, audience engagement, and the responsibility that comes with communicating science.

Attendees will walk away with a deeper understanding of how to produce compelling science communication content, the role digital media plays in democratizing science, and the immense potential this holds for education and society at large. Whether you're a content creator, a science communicator, or just an interested individual, this talk promises to be a game-changer.

Germany’s number 1 ranked chess player, Vincent Keymer, is taking on a double duty at the Big Techday 23. First, he will deliver a talk, explaining how modern chess has evolved over the past few decades due to computers and technology. Afterwards, Keymer will face a special challenge: simultaneous chess. He will compete against up to 20 guests and speakers of the Big Techday, providing ambitious amateur players with a rare opportunity to challenge a grandmaster. Keymer aims to demonstrate that it’s not just chess computers but also humans who are capable of exceptional performances at the board.

As the life insurance industry is moving towards investment-linked products, there is an increasing need for digitalization of the processes involved in these products, enabling customers to engage in a digital product experience over both web and mobile channels.

Munich Re, one of the world's leading reinsurers, has client relationships with many life insurance companies, through which it gains deep insights into the challenges and needs of different insurers worldwide. Its subsidiary Munich Re Markets GmbH was founded with the goal to support insurers by providing easily embeddable web components.

After several years of experimenting with different ideas, it turned out that full-stack components are the most promising approach to achieve this goal. These components, consisting of backend services and frontend components similar in spirit to microfrontends, provide a self-contained full-stack experience while at the same time integrating with business processes as well as the corporate design on the insurer's website. The implementation effort on the insurer's side is kept minimal by interfacing purely in the frontend and avoiding backend-to-backend integrations altogether.

Munich Re Markets has been working together with TNG and others to develop these components over more than 2 years. We present some of the unique challenges and obstacles we faced and the solutions we found during this process.

Do I actually know how to program? Some recent coding adventures have cast increasing doubts. I'll talk about some of these difficulties along with some big ideas that have changed everything I thought I knew about programming and problem solving. Can an old programmer learn new tricks? Maybe.

In this talk, Erwin shares his personal experience with drop-catching expired single-character domains the second they become available again. He takes the audience through the lifecycle of domain names, highlighting the unique characteristics of different top-level domains. He also provides detailed technical examples of snapped single-character domains, offering valuable insights for software developers.

This session discusses the neuroscience of developer productivity. It provides a thought-provoking take on the psychological root causes and “cognitive fatigue” associated with developer bottlenecks like unnecessarily long feedback cycles, protracted troubleshooting times, and toolchain failures. It discusses how a lack of understanding of this dynamic leads to a misalignment between leadership’s incremental perspective on the potential impact of productivity improvements and that of their developers who feel the cognitive pains every day and see the impact of addressing these pains as transformational. Translating the productivity benefits of removing cognitive fatigue pain into a hard business case for leadership consumption is the key to realizing the “DPE multiplier”, which is the number of times developers say they could be more effective with a more efficient toolchain (usually 2X-5X). The presentation concludes by making the connections between productivity, developer experience, and joy.

Picking up a new technology, developers stick with defaults. It's a fact that every tool from JQuery to NextJS has needed to face. At their worst, defaults ship hundreds of wasted JS kilobytes for routing, state, and other baggage developers may never use. But at their best, defaults give us a simple baseline to start from, with a clear path to add the exact set of features our project demands. This is the magic of opt-in design. Let's see how smart defaults guide modern frontend tools from Astro to React Server Components, and why this new era reshapes your workflow, and website performance metrics, for the better.

Are you building a microservice garden, or letting a jungle sprout? Microservice architectures are so efficient in software development because they take advantage of decentralization: giving choices to teams, enforcing clean factoring of code, and allowing more fine-grained operations. But if one is not careful an orderly microservice garden can turn into an untamed jungle. At Home Connect, BSH's smart home brand, we are addressing this problem with Backstage, an open source software catalog and developer platform which returns some carefully considered centralization back to the ecosystem. How does Backstage do this without interfering with developer autonomy? What else can it do? And might it be helpful in your project? That's what we will answer in this talk!

The era of exponential growth can be described as an age of rapid technological developments and an unprecedented increase in data and information. This development requires new ways of thinking!

But how do we deal with the reality that our brain is programmed to think linearly and that we have difficulties anticipating exponential developments?

Users are impatient. The loading time of a website is one of the fundamental aspects of a successful product. But not only the measurable waiting is important. Attractive websites also reduce the perceived waiting time.
In her talk, Cathrin will use examples from everyday project work to present how to avoid unnecessary errors when bootstrapping a JavaScript application. The following questions will be discussed:

• What can be done if a legacy backend prevents performance?

• Is configuration at runtime a reasonable requirement?

• How can the perceived waiting time be reduced with the help of modern CSS technologies?

At Rio, many independent teams work on a modern cloud-based logistics platform. In order to deliver features as quickly as possible and in high quality, we changed the development process to trunk based development a few years ago. Even highly critical services are developed without feature branches and deployed to production within 15 minutes several times a day. We would like to give an insight into our approach and highlight the most important techniques and challenges of trunk based development using a few examples.

This talk explores a series of surprising connections beginning with the simple question of two blocks colliding with each other, where an unexpected occurrence of pi pops up. Understanding what happens involves a tour through kinematics, optics, with a little willingness to dive into the rabbit holes that present themselves along the way, quantum computing.