Hi, I’m Pratyush Agnihotri 🙏
Researching AI. Developing Distributed Stream Processing. Impacting Systems. Educating the Next Generation.
I am a Deputy Head at Systems AI for Decision Support (SAIDE) research department, German Research Center for Artificial Intelligence (DFKI). At DFKI, I’m leading the federal ministry funded project etaGPT project. My work here focuses on designing and evaluating energy-efficient LLM architectures, especially for industrial environments. In addition, I am working as Postdoctoral Reseacher at Systems Group at TU Darmstadt. Previously, I completed my Ph.D. in ETiT department at TU Darmstadt, where my research focused on Distributed Stream Processing, AI-driven Optimization Models, and Efficient System Design for Heterogeneous Environments.
From June 2021 to 2024, I worked as a Research Scientist at the Multimedia Communication Lab (KOM) at TU Darmstadt, contributing primarily to the C2 subproject of the DFG Collaborative Research Center MAKI. Prior to Ph.D., I worked as a Software Developer at axxessio GmbH in Darmstadt, engaging in applied research and development in the fields of Internet of Things (IoT), Smart City Solutions, and Voice Assistants.
Before transitioning into academic research, I gained around ten years of industry and applied R&D experience across different domains, including Internet of Things (IoT), Smart City Solutions, Voice Assistants, connected systems, and industrial software systems. This included my last role as a Team Lead and Software Developer at axxessio GmbH in Darmstadt.
I began my academic journey by joining the Master’s program in Computer Science at TU Darmstadt in 2013. My Master’s thesis involved proposing and developing a decentralized crowd-sensing data collection framework based on Information-Centric Networking (ICN) to address post-disaster communication challenges. This work was conducted in close collaboration with Prof. Dr.-Ing. Michael Zink from the University of Massachusetts Amherst (USA).
Research Interests
- Sustainable AI-based Data Systems
- Distributed and Resource-efficient Stream Processing
- Autonomous Resource Management
- Performance Benchmarking and Optimization
- Heterogenous Cloud Environment
- Internet of Things (IoT)
Full CV: experience, education & academic service
Current Projects
I currently lead etaGPT, a federal ministry-funded project at DFKI building LLM-based data systems for energy data understanding, and work as a researcher on DFG MAKI Subproject C2, exploring transition-aware adaptation for event-based systems.
Recent Publication
Agnihotri, Pratyush, Luthra, Manisha and Binnig, Carsten. (2026). "EcoLLM: Energy-Aware Benchmarking of LLMs for Data ProcessingWorkloads." AIDM@SIGMOD 2026.
Agnihotri, Pratyush, Binnig, Carsten, and Luthra, Manisha. (2025). "Learning What Matters: Automatic Feature Selection for Learned Cost Models for Parallel Stream Processing." AIDB@VLDB 2025.
Agnihotri, Pratyush and Binnig, Carsten. (2025). "Demonstrating PDSP-Bench: A Benchmarking System for Parallel and Distributed Stream Processing." SIGMOD 2025.
Current Teaching Activities
I am actively involved in teaching activities focused on distributed systems, communication networks, and stream processing technologies. I have contributed as a Teaching Assistant for undergraduate and graduate courses, including “Communication Networks 2,” and regularly supervise Master’s theses, Bachelor’s theses, seminars, and lab projects. My teaching emphasizes hands-on experience, research-driven learning, and bridging theory with real-world system design. Read more
Open PhD, Theses and HiWi Positions
I am looking for motivated students who are interested to work on cutting edge technologies and IoT applications area. If you are interested to do PhD or write a Bachelor or Master Thesis or gain experience in my current ongoing projects as HiWi then please feel free to contact me. Send your CV and Transcript list on my email address.