Research Assistant in Security of Mechanistic Interpretability (AMIRA project)
- Deadline
- 26 Nov 2026 · 4 days left Closing soon
- Location
- Berlin, Germany
- Type
- Research Assistant
- Compensation
- TV-L Berliner Hochschulen E13
- Duration
- 36 months · 3 years
- Working language
- English
- Posted
- 19 Nov 2026
- Verified
- 19 Nov 2026
About this opportunity
The Chair of Machine Learning and Security at Technische Universität Berlin (Faculty IV, Electrical Engineering and Computer Science, and BIFOLD) is hiring a research assistant for the AMIRA project. The project studies the security of mechanistic interpretability methods for machine learning models. This is a three-year, fully paid position on the TV-L 13 scale, and you can work towards a PhD during it.
What is offered
- Paid position at salary grade 13 TV-L Berliner Hochschulen
- Fixed-term contract for 3 years, starting as soon as possible
- Full-time, though part-time may be possible
- The chance to work towards a doctorate (PhD) during the job
- Based in Berlin-Charlottenburg, inside the BIFOLD research environment
Your main tasks would be:
- Analysing how secure mechanistic interpretability methods are
- Building attacks against these methods and defences for them
- Publishing your results in scientific venues
Who can apply
- You need a completed Master's, Diplom or equivalent degree in computer science or a related technical field
- You should have a strong background in machine learning and computer security
- Experience with adversarial machine learning and mechanistic interpretability is expected
- The ability to do research on your own is a plus
- You need good German and/or English. If you only speak one of them, you should be willing to learn the other
How to apply
- Email your application as one PDF file (max. 5 MB) to hashmi@tu-berlin.de
- Put the reference number IV-428/26 in your application
- Include a cover letter, your CV, and the certificates and transcripts for both your Master's and Bachelor's degrees
- Recommendation letters and a list of publications are optional
- Deadline: 26 November 2026
- Contact: Prof. Dr. Rieck
Requirements
-
Open to all nationalities
-
PhD candidate
-
Minimum degree: Master's
Research Assistant in Security of Mechanistic Interpretability (AMIRA project)
26 Nov 2026 · 4 days left