Asterion Labs06 · Asterion Labs
We hire researchers and engineers who are comfortable being wrong quickly. Every role here is measured by how much uncertainty about the universe it removes.
R-01
Discovery Loop
Zürich · Hybrid
Own the systems that turn generated mathematics into executable experiments. You will work between the symbolic layer and the simulation scheduler, and your code will decide which theories get compute.
R-02
Hypothesis Generation
Zürich / Pasadena
Improve the models that propose novel field equations. This is generative modelling over formal mathematical structures, evaluated by physics rather than by human preference.
R-03
Observational Validation
Remote · EU / US
Design the confrontation between machine-proposed theory and real sky data. You will define likelihoods, systematics models and blind-analysis protocols across CMB, lensing and BAO datasets.
R-04
Distributed Compute
Pasadena · On-site
Run a six-thousand-accelerator fleet as a market for scientific information. Scheduling, failure domains, reproducibility guarantees and the observability that makes results auditable.
No timed puzzles. Candidates review a real derivation graph produced by the engine and tell us where it is wrong. We pay for the exercise.