Principal Bayesian Policy Inference Architect

corpora-jobs-1778796293285-db9d41c6 - Frontier Development

Research ScientistPrincipal1 position

⚡

Why This Role is Different

Frontier Development Role

Architect the probabilistic inference backbone that lets multi‑agent systems reason about policy uncertainty in the presence of adversarial observation noise, enabling autonomous decision‑making under threat.

The Frontier Element

Your work will pioneer a joint GAN‑Bayesian inference pipeline that can be queried in real time on distributed agents, a novel contribution to robust MARL that blends generative modeling with hierarchical Bayesian inference.

🔬

Project Context

Research Area

Bayesian Policy Inference (BPI) with marginalization over generative observation model

From: Adversarial Observation Perturbations and Policy Inference

Why This Role is Critical

Essential to build the hierarchical Bayesian framework that produces robust policy posteriors under unseen adversarial perturbations.

What You Will Build

Amortized variational inference engine, Monte Carlo integration module, policy prior integration with CRL, and a scalable inference runtime for distributed agents.

🛠

Key Responsibilities

Design the hierarchical Bayesian model that treats policies as latent variables and marginalizes over the CC‑GAN observation likelihood.
Implement amortized variational inference with efficient Monte Carlo sampling for real‑time policy posterior estimation.
Integrate the Bayesian posterior into the cooperative resilience layer, enabling adaptive recovery policies.
Benchmark robustness against unseen adversarial perturbations on simulated UAV and robotic swarm environments.
Collaborate with the Generative Observation Modeling team to co‑train the generative and inference components.

🎯

Required Skills & Experience

Technical Must-Haves

Hierarchical Bayesian modeling and inference

Expert

Core to marginalizing over observation uncertainty.

Amortized variational inference (e.g., VAE, IWAE) and Monte Carlo integration

Advanced

Needed for scalable, real‑time inference.

Reinforcement learning (MARL, CTDE, MAPPO)

Advanced

Understanding policy training and integration with Bayesian inference.

Probabilistic programming (Pyro, TensorFlow Probability)

Proficient

Facilitates rapid prototyping of complex Bayesian models.

Python, C++, CUDA for high‑performance inference

Expert

Required for production‑grade inference runtime.

Experience Requirements

7+ years in Bayesian machine learning or probabilistic robotics.
Published work on Bayesian inference in RL or adversarial settings.
Experience deploying inference engines on distributed hardware.

Education

PhD in Statistics, Machine Learning, or Robotics with a focus on Bayesian methods.

⭐

Preferred Skills

Knowledge of adversarial variational inference and domain adaptation.
Experience with counterfactual explainability in Bayesian models.
Background in quantum‑enhanced inference or digital twin concepts.

🤝

You Will Thrive Here If...

Thrives in environments where theory meets rapid prototyping.
Comfortable owning end‑to‑end systems from model to deployment.
Eager to experiment with novel inference architectures.

📈

Impact & Growth

12-Month Impact

Achieve a 25% improvement in cooperative task success under adversarial telemetry in simulated UAV swarms within 12 months, demonstrating the practical value of Bayesian policy inference.

Growth Opportunity

Scale the inference engine to multi‑domain deployments (e.g., autonomous driving, cyber‑defense) and lead the integration of counterfactual explainability for human‑in‑the‑loop oversight.

Ready to Push the Boundaries?

If this sounds like the challenge you have been looking for, we want to hear from you. We value what you can build over where you have been.