Master Thesis in Continual Learning with Agentic Memories

At Bosch, we shape the future by inventing high-quality technologies and services that spark enthusiasm and enrich people’s lives. Our promise to our associates is rock-solid: we grow together, we enjoy our work, and we inspire each other. Join in and feel the difference.

The Robert Bosch GmbH is looking forward to your application!

Are you looking for an opportunity to explore how machine learning systems can continuously evolve while maintaining up to date knowledge? In this thesis, you will dive into the challenges of agentic systems, where growing context sizes drive up training costs, and focus on improving their working memory and token efficiency while applying common LLM models in continually learning and practice-oriented settings.

• You will begin your thesis by conducting a comprehensive literature review on memory in agents, analyzing existing benchmark implementations, datasets, and methods to build a deep understanding of the field while also exploring the domain of continual learning.
• Building on this foundation, you will adapt existing benchmarks or implement your own for Bosch-related use cases. In this context, you will write code to apply LLMs in an agentic setting, with a particular focus on agent memory.
• Based on these insights, you will derive and implement methods aimed at improving the memory of continually learning agentic systems.
• Finally, you will rigorously evaluate the performance of the developed approaches on standard academic benchmarks as well as Bosch use cases, while you will analyze scalability, robustness, and deployment potential.
• You will carry out all of these tasks within a tight project timeline, with your results strongly encouraged to be submitted to major upcoming machine learning conferences, while performing effectively under deadline driven time pressure is mandatory.

• Education: master studies in the field of Computer Science, Mathematics, Machine Learning or comparable with a focus on machine learning with very good grades
• Experience and Knowledge: 
  • strong academic background in machine learning and (applied) mathematics
  • solid programming skills in deep learning with PyTorch as well as proficiency in Git
  • familiarity with job scheduling systems
  • practical knowledge of agentic systems and their implementation in a research setting
  • background in working with LLMs using PyTorch and Python
• Personality and Working Practice: you are a motivated and research oriented person who takes a proactive and independent approach to problem solving and is able to work effectively under deadline pressure
• Work Routine: our hybrid model provides you with a balanced mix of on site presence and remote work (70% remote, 30% in presence)
• Enthusiasm: keen interest in independent problem solving
• Languages: fluent in English and beginner in German

Start: according to prior agreement

Duration: 6 months

Requirement for this thesis is the enrollment at university. Please attach your CV, transcript of records, examination regulations, list of previous code projects (need not be published) with brief descriptions and if indicated a valid work and residence permit.

Diversity and inclusion are not just trends for us but are firmly anchored in our corporate culture. Therefore, we welcome all applications, regardless of gender, age, disability, religion, ethnic origin or sexual identity.

Need further information about the job?

Pascal Janetzky (Functional Department)

+49 173 4163104

Michael Klar (Functional Department)

+49 1525 8813540

Work #LikeABosch starts here: Apply now!

#LI-DNI