Manisha Luthra Agnihotri wants to develop the next generation of stream processing systems that process continuous data at lightning speed. What sets her vision apart is the integration of artificial intelligence: her systems are designed to adapt on the fly to new data types such as text, images, or audio, and to dynamic scenarios in fields like healthcare, finance, and robotics.

Originally from India, Manisha Luthra Agnihotri has a bold vision: to develop intelligent stream processing systems that can interpret and act on real-time data – not just numbers, but also language, images, even audio. These systems will do more than just compute fast – they will learn, adapt, and reason as data flows in, powering a new generation of applications in healthcare, robotics, and autonomous systems.

This is no small task. “Streaming systems today are used in time-sensitive environments. But they are limited to conventional data and fixed processing logic,” explains Luthra Agnihotri. “I want to change that – to build AI-based systems that can adapt to new modalities and unforeseen events automatically.”

Research at TU Darmstadt and DFKI

To make this possible, her lab at TU Darmstadt and at the German Research Center for Artificial Intelligence (DFKI) is designing one of the first multimodal stream processing engines. This next-generation system can understand live data – whether text from patient records, speech from a robot’s microphone, or sensor output from autonomous vehicles – and react immediately. At the heart of this innovation are large language models (LLMs) like GPT, which provide new capabilities but also introduce deep technical challenges. “We need to rethink everything – how queries are processed, how we manage latency, how we ensure reliability,” she says.

These models open up new possibilities – such as understanding medical reports, camera feeds, or spoken commands as they arrive – but they also bring with them complex technical challenges. “How do we maintain speed when using such heavy models? How can we ensure correctness and safety in autonomous settings? These are the types of questions we’re tackling,” she says.

Her work is highly interdisciplinary and collaborative. At DFKI, she works closely with Professors Jan Peters from the robotics area and Kristian Kersting from the core machine learning area. “Their expertise in decision-making and learning from multimodal data enriches the core of my systems research,” she says. Together, they are exploring how AI models can process real-time signals and translate them into robust actions in robotic and autonomous systems. Her research opens the door to a whole new class of applications: real-time decision support for medical professionals, robotic systems that understand multimodal input on the go, or AI-driven financial tools that can act on live text and image data.

From New Delhi to Darmstadt

The 34-year-old computer scientist from New Delhi has come a long way since arriving in Darmstadt in 2013 for her Master’s degree. She had several international offers – including from Australia and other European countries – but was drawn to Germany’s academic structure and TU Darmstadt’s curriculum.

During her studies, she became involved in the Collaborative Research Center MAKI (“Multi-Mechanism Adaptation for the Future Internet”), where she worked on distributed systems and the fusion of different data modalities – a topic that has become central to her work today. Her early contributions were recognized with the MAKI Female Student Travel Award and the MAKI Networking Award for her outstanding master thesis. After earning her degree with distinction, she stayed to pursue a PhD, turning down offers from the USA, UK, and Stuttgart. “Darmstadt simply had the strongest research profile in the areas I cared about – and it won me over,” she says.

Her doctoral work focused on building adaptive stream processing systems and data networks in the MAKI Collaborative Research Center. Even then, she was exploring new methods for integrating multiple data modalities – a challenge that has only grown more relevant with the rise of AI. Her PhD was awarded summa cum laude and received the national KuVS prize for the best doctoral thesis in distributed systems.

Today, Luthra Agnihotri is a postdoc in the Systems Group at TU Darmstadt and deputy head of research in the “Systemic AI for Decision Support” group at the German Research Center for Artificial Intelligence (DFKI), led together with Professor Carsten Binnig. The combination of academic and applied research suits her perfectly. “It allows me to shape future systems while staying grounded in real-world needs,” she says.

Athene Young Investigator as a stepping stone

The Athene Young Investigator award recognizes her achievements – and supports what comes next. “It gives me independence – the freedom to take risks, build a team, supervise PhD students, and run my own research agenda,” she says. That work has already begun: Luthra Agnihotri is currently supervising two doctoral students, and a third PhD joined her lab this year.

For her, the award also comes with a broader mission: to mentor the next generation, especially women in computer science. “I want to show that this space belongs to them too.”

Darmstadt has become home in more ways than one. She recently had her first child and finds the city warm, international, and intellectually vibrant. “I had other options,” she says with a smile, “but I chose to stay. It has given me so much – not just professionally, but personally too.”

Dr Manisha Luthra Agnihotri, postdoctoral researcher in the Systems Group at the Department of Computer Science and deputy head of research in the department of ‘Systems AI for Decision Support’ at the German Research Center for Artificial Intelligence (DFKI)

Text: TU Darmstadt

How can carbon-free chemical fuels like hydrogen, ammonia, and metal powders be used as energy carriers for a sustainable future? Dr. Tao Li, newly appointed Athene Young Investigator, focuses his research on the environmentally friendly and energy-efficient use of ammonia. The mechanical engineer at the Department of Reactive Flows and Diagnostics relies on plasma – still a relatively new method – to unlock ammonia’s potential as an alternative fuel.

Most people associate the pungent-smelling substance with cleaning agents or fertilizer. It can also be found in human urine. The chemical compound, consisting of hydrogen and nitrogen, is widely used in industry but its production is still energy intensive. Could it be produced more sustainably and used as a fuel for the future? “Ammonia is mainly produced industrially using the Haber-Bosch process,” explains Dr. Tao Li. This method combines molecular nitrogen from the air with molecular hydrogen using catalysts. “However, it requires high temperatures of around 500°C and pressures of around 200 bar,” says the mechanical engineer. Both factors consume large amounts of energy and drive up costs. As a result, large-scale facilities with substantial capital investment are required.

Tao Li is taking a different approach: plasma-driven catalytic synthesis – an emerging concept that researchers have only been exploring in the recent years. “Ammonia produced from green hydrogen offers a carbon-free fuel alternative,” the 37-year-old explains. The challenge lies in its low reactivity, which currently limits practical applications. Li aims to enhance this reactivity through the use of plasma. During ammonia synthesis, plasma is generated under high voltage. It acts as a kind of booster, enabling the conversion of hydrogen and nitrogen into ammonia even at room temperature and atmospheric pressure. Additionally, plasma can enhance energy release by promoting more efficient combustion of ammonia. A kinetic enhancement effect is observed in plasma-assisted combustion, where plasma accelerates ammonia oxidation and enables better control of NOₓ emissions. “My research focuses on non-equilibrium plasma to improve ammonia synthesis and combustion with minimal energy input. Our group has developed plasma reactors that have high potential to enhance both synthesis efficiency and combustion performance,” says the AYI researcher.

Energy-efficient ammonia synthesis

This method could make alternative fuels more accessible, requiring fewer devices while reducing costs and environmental impact. According to Li, ammonia is easier to liquefy and transport than hydrogen and does not pose the same explosion risks. It could be a viable alternative fuel for industry, aviation engines, or even households. “Fuels like ammonia can store intermittent renewable energy from solar and wind over long periods and be transported globally. This addresses key challenges of renewables – fluctuations and geographical limitations. By replacing fossil fuels in power generation, transportation, and industrial heating, these green chemicals are vital for sustainable and affordable decarbonization,” he emphasizes.

Tao Li's research also delves into the fundamental energy conversion mechanisms of these fuels in thermochemical processes. Unlike fossil fuels, green alternatives exhibit radically different combustion characteristics – such as hydrogen's explosiveness, ammonia's low reactivity, or particulate emissions from metal fuels. “Adapting existing infrastructure requires a deep understanding of their complex, multiphase reaction dynamics involving chemical kinetics, fluid mechanics, and heat and mass transfer.” To investigate these physical processes, he is developing advanced multi-physics laser diagnostics that enable real-time measurements of critical parameters such as temperature, species concentrations, and particle dynamics. “These data not only shed light on the underlying science, but also directly contribute to the design of next-generation reactors and the optimization of industrial processes,” he explains. He is convinced that by linking fundamental research to technical applications, his work can accelerate the implementation of carbon-neutral energy solutions.

Privilege to conduct research at TU Darmstadt

The TU Darmstadt offers the scientist – born in Tangshan, a mid-sized city in China –, “an ideal environment for cutting-edge research.” Tao Li completed his bachelor’s degree in mechanical engineering at Tongji University in Shanghai, a partner university of TU Darmstadt. He chose to move to Germany for his master’s and later for his doctorate, both completed in Darmstadt. “It is a privilege to conduct research at one of the world’s leading laboratories for advanced laser diagnostics and combustion, and to work with outstanding scientists on topics of great economic and societal relevance,” he says. Receiving the AYI title is, for the father of a young family, “a great honor and meaningful recognition of the potential of my research to drive sustainable energy innovation.”

The award highlights the highly interdisciplinary nature of his work, which combines experimental and numerical methods from combustion science, chemistry, and materials science. The AYI supports him in building his own research group, supervising PhD students, and pursuing independent research directions with greater freedom. Tao Li believes the prestige of the award will also be crucial for establishing new collaborations with other institutions and industrial partners. At the same time, he hopes the recognition will help advance his academic career: the young researcher aspires to become a professor.

Dr.-Ing. Tao Li, Research Group Leader at the Institute of Reactive Flows and Diagnostics, Department of Mechanical Engineering

Autorin: Astrid Ludwig

Initially, Philipp Rosendahl did not have an academic career in mind. But then his enthusiasm for ski touring and mountains became a passion, and his doctoral thesis in mechanical engineering evolved into the Center for Snow and Avalanche Research at the Department of Civil and Environmental Engineering – unique at TU Darmstadt. The 35-year-old has now been appointed Athene Young Investigator.

When the native Berliner puts on his skis for the next mountain tour, he always has the most important companions in his luggage: his avalanche backpack – equipped with an airbag system and an avalanche transceiver, or beeper for short. This enables him to locate avalanche victims or, in an emergency, to be located himself. "Avalanches pose a real danger to ski tourers," says Dr Philipp Rosendahl. Even before becoming a dedicated avalanche researcher, the keen skier had already studied the subject intensively, always checking the weather and snow reports before going on a tour.

But how does a mechanical engineer become a nivologist – a snow researcher? "It was pure interest," says the 35-year-old. Rosendahl completed his bachelor's and master's degrees at the Technical University of Darmstadt, followed by a doctorate. He conducted research on fracture mechanics under Professor Wilfried Becker focusing particularly on adhesive joints like those used in car manufacturing to protect passengers by absorbing kinetic energy in the event of an impact. As a doctoral student, he investigated, among other things, the mechanical behaviour and how cracks form. In the process, the winter sports enthusiast realised that adhesive joints and layers of snow are similar in many ways. "One of these layers will be the weakest and break first." Often it is the surface hoarfrost, a thin, fragile layer of fine ice crystals covered by freshly fallen snow. "The mathematical and mechanical principles are essentially the same."

Initial contact with the research community

In a mechanical engineering project he supervised, Rosendahl looked at how this knowledge could be used in snow research and to improve avalanche warnings. In 2018, this work led him to the International Snow Science Workshop (ISSW) in Innsbruck – a major international conference for snow and avalanche researchers and practitioners held every two years in Europe, the US or Canada. "I was immediately well received there and there was a lot of interest in working together," he reported. Rosendahl's topic filled a gap: "The engineering perspective had been missing from this community." There are many researchers from physics, meteorology or geosciences there, but no mechanical engineers. This was a new impetus from outside, and it also benefited the Darmstadt engineer himself. "I met passionate outdoor people, a small, warm research community without the usual scientific competition."

This gave him access to databases, research facilities and collaborations, such as with the Swiss Federal Institute for Snow and Avalanche Research in Davos. He learned to differentiate between continental and maritime snow types and their unique characteristics. "Snow research has many aspects – from hydrology to climate change, and the increasingly pressing question of what actually happens as temperatures rise." In field experiments and in exchange with practitioners, he created snow profiles and researched the icy masses layer by layer.

After successfully completing his doctorate at the Department of Mechanical Engineering, it was clear to the young researcher that "I wanted to stay at the university as a postdoc, though this was never my original plan." Dr Philipp Rosendahl found a kindred spirit in Professor Jens Schneider – at the time head of the Institute of Structural Mechanics and Design at the Department of Civil and Environmental Engineering, and now rector of TU Wien. "He gave me the opportunity to do snow research."

Establishment of the Center of Snow and Avalanche Research

In 2020, the Center of Snow and Avalanche Research was established at the Technical University of Darmstadt, where the 35-year-old is now a group leader and for which he has now been selected by the TU as an Athene Young Investigator. The university uses the funding to support the academic careers of its young researchers. The center’s work is unique. Rosendahl and his team create computational models and devise experiments using mathematical and numerical methods from fracture mechanics. Field tests are conducted at the Swiss Federal Institute for Snow and Avalanche Research in Davos. "They complement each other well." The technical expertise from Darmstadt has also led to the development of completely new devices, such as a mobile snow tester, designed for backpacks by TU doctoral student Valentin Adam and TU master's student Luis Berger. Rosendahl emphasizes its significance: "Logistical challenges in mountainous snow regions are always substantial." Being a snow researcher is physically demanding. "You have to be fit."

Rosendahl is currently working with two doctoral students. More will join them by the end of 2025. "We have a lot of research proposals in the pipeline." In 2023, Philipp Rosendahl was awarded TU Darmstadt's Dr. Hans Messer Stiftungspreis, an award recognizing young researchers in early career stages. He is particularly pleased to receive the award as an Athene Young Investigator. "It gives me a whole new perception and a different standing. Previously, I was a postdoc; now I supervise doctoral students and carry out professor-level tasks." It's a big change that gives him more access to third-party funding and more freedom in research. Perhaps it will also help the snow expert fulfil his dream of conducting research north of the Arctic Circle or in the Himalayas.

Dr.-Ing. Philipp Rosendahl, Group Head of the Center of Snow and Avanlanche Research, Institute of Structural Mechanics and Design at the Department of Civil Bau- und Unweltingeineurwissenschaften

Autorin: Astrid Ludwig