Table of Contents

Seminar with Practical: Scalable Computing Systems and Applications in AI, Big Data and HPC

Key information

Contact Julian Kunkel, Jonathan Decker
Location Virtual
Time Thursday 14:15-15:45
Language English or German (individual presentation)
Module M.Inf.1238: Scalable Computing Systems and Applications in AI, BigData and HPC
SWS 3
Credits 5
Contact time 42 hours
Independent study 108 hours

As part of this seminar, you will create a presentation, work on a small-scale practical project and write a report revolving around a research topic in German or English (your choice!). Therefore, you will meet regularly with an assigned supervisor and work towards the presentation, practical project and report. You will first select a topic and a use case related to the overall topic of the course. Then, during the term you will prepare a presentation to introduce the topic and the state of the art. Next, you will realize a small-scale project by practically working on your topic. This includes evaluating performance and scalability, as well as analyzing and quantifying the contribution of your topic or tool. Finally, you present your results in another presentation.

The presentation time is 25 minutes (plus discussion) for each presentation. A short report describing your work in the practical project is expected (max 15 pages).

Please note that we plan to record sessions (lectures and seminar talks) with the intent of providing the recordings via BBB to other students but also to publish and link the recordings on YouTube for future terms. If you appear in any of the recordings via voice, camera or screen share, we need your consent to publish the recordings. See also this Slide.

Required Prior Knowledge

Learning Objectives

By the end of the seminar, students can:

Topics

This is the list of topics that we will assign to students during the first meeting. You will have some room for developing the topic into the direction of your choice. Feel free to propose your own great topic.

Examination

The exam is conducted as part of the final presentation (30% of the mark) and the report (70%).

The assignment sheet with detailed expectations is available here: Assignment sheet.

Agenda

Topic Distribution

Student Supervisor Topic Submissions
Your Name Your Supervisor Your Topic Report
Moritz Kuschel Jonathan Decker How can knowledge graphs be applied as effective context organization tool to provide task-specific meaning for humans or agents?
Janik Hartger Michael B.Khani How do risk-adaptive authorization and deployment strategies affect the security, task completion rate, and performance of autonomous AI agents in service environments?
Laura-Sophie Kanne Mojtaba Akbari Scalable Databases
Nirish Samant Anila Ghazanfar Coding Agents from First Principles
Luis Willem Kretzschmar Sepehr Mahmoodian Building an End-to-End Edge-to-HPC Streaming Pipeline
Mudit Tandon Aasish Sharma Using LLMs to Generate Knowledge Bases
Mohd Uwaish Lauritz Rasbach Benchmarking Sparse Attention in LLMs
Elias Lautensack Jonathan Decker Coding Agents from First Principles
Mohamed Najem Mojtaba Akbari What's new in the Kubernetes ecosystem
Amin Niaziardekani Mojtaba Akbari Service Discovery and Traffic Management in Cloud Applications
Naveen Reddy Patrick Höhn Limitations of AI Agents in Coding
Colak Naz Lauritz Rasbach Benchmarking Sparse Attention in LLMs
Arthur Lenz Patrick Höhn Limitations of AI Agents in Coding
Philipp-Basjan Sibma Patrick Höhn Limitations of AI Agents in Coding
Ugur Can Boyraz Mojtaba Akbari Scalable databases with e.g., Elasticsearch, Postgres
Selma Lafi Aasish Sharma Using Large Language Models to Generate Knowledge Base
Raj Randive Sepehr Mahmoodian Benchmarking Streaming Systems for HPC Data Ingestion
Vaishnavi Dharmeshbhai Lalwala Lauritz Rasbach Benchmarking Sparse Attention in LLMs
Maxim Barnstorf Sepehr Mahmoodian Building an End-to-End Edge-to-HPC Streaming Pipeline
Aayush Gupta Patrick Höhn Application of agentic coding to maintenance of scientific software
Ojaswi Ojaswi Sepehr Mahmoodian Building an End-to-End Edge-to-HPC Streaming Pipeline