Industry Thesis - GenAI for Microservice Performance Modeling & Analysis
| Typ | Masterarbeit | |
|---|---|---|
| Aushang | ||
| Betreuer | Wenden Sie sich bei Interesse oder Fragen bitte an: Maximilian Hummel (E-Mail: maximilian.hummel@kit.edu) |
Motivation & Problem Statement
Modern industrial software systems are increasingly built as microservice architectures. While this enables flexibility and rapid evolution, it also creates major challenges for understanding and managing performance behavior across many interacting services. Small changes in one component can propagate through the system and cause unexpected bottlenecks, latency spikes, or reduced reliability under load.
At ABB’s Corporate Research Center in Mannheim, we investigate how Generative AI can support software architecture to make such complex systems more understandable and actionable. For example, a key open problem is identifying which parts of a microservice landscape are most critical for end-to-end performance outcomes. Today, this assessment is often manual, time-consuming, and hard to scale during early design.
This thesis addresses that gap by developing a criticality analyzer within an existing GenAI-based performance modeling project. The analyzer should help stakeholders prioritize where to focus optimization, validation, and architectural improvements. The overarching motivation is to reduce modeling effort, increase transparency, and improve decision quality for performance engineering in complex distributed systems.
Planned Contribution
The thesis will contribute a concept and implementation of a criticality analyzer that fits into the current project ecosystem for GenAI-assisted microservice performance modeling.
Planned outcomes include:
- A clear definition of “criticality” in the context of microservice performance modeling.
- A analysis workflow that uses available model/context information to estimate component importance.
- A structured qualitative/quantitative assessment on selected example scenarios (ABB and open source) to demonstrate usefulness.
ABB Corporate Research Center
The ABB Corporate Research Center in Germany is one of ABB’s global research centers with about 90 innovative scientists and engineers from 15+ different countries and with around a dozen of technical disciplines. Our two departments, "Control Technology & Automation Components" and "Digitalization & Software Technologies", devise cutting-edge new technologies for the transformation of production and energy systems. The thesis is hosted at the ABB Corporate Research Center in Mannheim, with direct industry contact and a modern workplace. The setup can be hybrid (on-site + remote by agreement). In addition, ABB provides financial compensation during the thesis period and there is an impact-based bonus program: higher bonus levels are granted when thesis results create stronger measurable value for ABB research and innovation.
Requirements
- Solid programming skills in Python and confidence working with existing codebases.
- Basic understanding of software architecture and distributed/microservice systems.
- Interest in performance engineering, model-based approaches, and applied AI/GenAI.
- Willingness to collaborate with ABB researchers on-site in Mannheim (hybrid arrangement possible by agreement).
Nice-to-have (not mandatory):
- Prior exposure to performance analysis, simulation/modeling (PCM), or software quality metrics.
- Experience with LLMs especially frameworks like LangGraph