Open-weight model safety
What actually survives fine-tuning, quantization, and open release, once the deployment-time guardrails are gone.
AI Safety Researcher · ML Engineer
Andreas Hermann
I'm an ML engineer and researcher moving into AI safety full-time. I did a PhD, then spent three years taking machine-learning systems into production. Now I work on two things: keeping open-weight models safe after they're released, and the misalignment that shows up when you wire models together into agents.
Now
Right now I'm moving out of applied ML leadership and into independent AI safety research. From August 2026 it's full-time, funded by a research transition grant.
The bet
I spent years keeping machine-learning systems running in production. That's what shaped how I think about safety. I've watched which safety properties hold up once a model leaves the lab, and which ones quietly fall apart, and that gap is the thing I want to work on.
There's real research behind it too. I did a PhD, published 17 peer-reviewed papers (450+ citations), and spent the last year properly upskilling through ARENA, the AI Alignment Research Fellowship, and BlueDot. Read the longer version →
Career
-
2026 – 2027
Independent AI Safety Researcher
Full-time research on open-weight model safety, compositional misalignment, and interpretability, backed by a BlueDot Impact transition grant.
-
2026
AI Safety Research Fellow · Safe AI Germany (SAIGE)
Inoculation against model poisoning: training a model so its safety properties can't just be fine-tuned away later.
-
2026
Facilitator · BlueDot Impact
Teaching technical AI safety cohorts and contributing to the alignment curriculum.
-
– 2026
ML / AI Engineering Lead
Three years taking machine-learning systems into production at scale, across classification, attribute mapping, intake, and data-quality AI.
-
PhD
Doctorate · University of Münster (WWU)
Information systems. REACH Award for the dissertation.
Research focus
Compositional misalignment
Why alignment tested on single models fails to compose once you chain them into multi-agent orchestrations, tool chains, and memory-augmented agents.
Interpretability & evaluation
The measurement tooling for safety: probes, evals, and mechanistic analysis that hold up in a real deployment.
Selected publications
-
2026
Differences in Alignment Behavior between Single-Agent and Multi-Agent LLM Systems
Recent writing
- 2026-06
- 2026-05
- 2026-04
Reading list
Curated papers, courses, and tools annotated for AI safety researchers and engineers crossing over from adjacent fields.