Founding ML Engineer
Turn cognition models, populations and psychological mechanisms into heterogeneous synthetic behavior.
This is not primarily a conventional generative-AI application role. Our researchers help define how people behave. You help make those behaviors computationally real.
What you would own
THYMOS and simulation architecture
Agent state, memory and behavioral dynamics
Probabilistic variation and population generation
Synthetic audience construction and agent heterogeneity
Model routing, uncertainty and behavioral calibration
Evaluation, reproducibility and large-scale inference
ML infrastructure, performance and cost optimization
Problems you might solve
Prevent thousands of agents from becoming reworded versions of one model
Give past experiences meaningful influence on future decisions
Calibrate simulated populations against empirical distributions
Propagate uncertainty instead of collapsing into a confident answer
Run large populations efficiently while retaining individual differences
Translate trust, cognitive load, hesitation and memory decay into computational behavior
Technical profile
Python and PyTorch, JAX or equivalent frameworks
LLM APIs, open-weight models and inference systems
Embeddings, probabilistic modeling and agent architectures
Evaluation systems, distributed or async processing and APIs
Databases, experiment tracking and statistical analysis
Rust, behavioral modeling, RL, multi-agent systems or simulation are valuable advantages
What this role requires
You enjoy ambiguous technical problems. You can move from research paper to hypothesis, prototype, experiment, evaluation and production system. You are comfortable challenging assumptions when evidence says they are wrong.
We are looking for consistent, serious involvement as part of Aetherya’s core technical team, not occasional contribution.
Compensation
This is an equity-compensated founding-team role. Indicative grant: approximately 1.0% of Aetherya, based on ability, commitment, responsibility, opportunity cost and contribution to core IP. Vesting and final legal documentation apply. There is currently no cash salary.
What success looks like
A substantially improved THYMOS architecture
Calibrated cognition models
Large-scale population simulation infrastructure
Robust evaluation systems and persistent agent state
Improved model economics and published technical research
Apply for this role.We read every application ourselves.
Aetherya is a frontier cognitive simulation lab building computational models of human perception, behavior and decision-making.
Our research focuses on synthetic humans: persistent, behaviorally grounded agents capable of interpreting information, experiencing uncertainty, forming objections, changing beliefs and making decisions within simulated environments.