Julian Schulz

AI Safety Researcher & Teacher

Research Experience

PIRAMID, Principles of IntelligenceResearcher (half-time)
2026 – present
Working with Andrew Mack on applying ambitious interpretability methods at scale to AI safety use cases.
Meridian Visiting Researcher ProgrammeAI Safety Researcher
2025 – present
Research on LLMs learning encoded reasoning during RL. Funded by a Coefficient Giving grant.
Open Philanthropy GrantIndependent AI Safety Researcher
2024
Research on Sleeper Agents and Automated Feature Labeling (ICLR 2025 workshop).
MATSScholar, mentored by Alex Turner
2023
Research on activation steering.

Publications

Julian Schulz, Lukas Fülle, Rieke Fruengel·NeurIPS 2026 Workshop AI4GOOD
Julian Schulz·COLM 2026 Workshop AdvML-Frontiers x CoTMA
Jou Barzdukas, Jack Peck, Julian Schulz, Paulius Rauba, Steven Basart, Lennie Wells
Julian Schulz, Seamus Fallows·ICLR 2025 Workshop Building Trust
Nina Panickssery, Nick Gabrieli, Julian Schulz, Meg Tong, Evan Hubinger, Alexander Matt Turner·ACL 2024

Teaching & Mentoring

ML4GoodHead Teacher
2025 – present
Teaching and leading the teaching team at week-long AI safety bootcamps in Europe, Canada, Singapore and South Africa.
Iliad IntensiveTeacher
2026 – present
Teaching Mechanistic Interpretability and Intro to ML Engineering.
MARSResearch Mentor
2025, 2026
Mentored research teams in two cohorts of Cambridge’s alignment research mentorship programme.
SAIGE IncubatorResearch Mentor
2026
Mentored the team behind the steganography paper above (NeurIPS 2026 workshop, oral).

Industry

Alexander Thamm GmbHMachine Learning Engineer
2024 – 2025
Developing and deploying multi-agent systems; managing cloud infrastructure with Terraform.

Education

M.Sc. Biophysics and Complex SystemsMax Planck School Matter to Life, full scholarshipGrade 1.6
2024
B.A. Philosophy & PhysicsGöttingenGrade 1.3
2022
B.Sc. PhysicsGöttingenGrade 1.5
2021
Studienstiftung des deutschen VolkesMerit scholarship for fewer than 0.5% of German students
2018 – 2024
ARENAAlignment Research Engineer Accelerator
2024
German grades: 1.0 is best.

Skills

Machine LearningPyTorch, RL (TRL, Tinker), W&B, Mechanistic Interpretability
EngineeringPython (5+ years), Terraform, Azure