Principal Causal Discovery & Bayesian Graph Learning Engineer

corpora-jobs-1778796293285-db9d41c6 - Frontier Development

Research ScientistPrincipal1 position

⚡

Why This Role is Different

Frontier Development Role

Lead the frontier of causal inference in high‑stakes multi‑agent systems. You’ll design algorithms that turn noisy, partially observable logs into a principled causal fabric, enabling trustworthy blame signals that survive adversarial manipulation.

The Frontier Element

You’ll pioneer hybrid Bayesian‑neural causal discovery that blends PC/NOTEARS with graph‑neural‑network priors, achieving online, cycle‑aware learning in non‑stationary environments—a capability that has no commercial precedent.

🔬

Project Context

Research Area

Causal Discovery Layer of CRAN

From: Misattribution of Blame in Cooperative Multi‑Agent Systems

Why This Role is Critical

The causal graph is the foundation of blame attribution; without a robust, online‑learning causal model, the entire CRAN framework collapses.

What You Will Build

An end‑to‑end causal discovery engine that learns Bayesian DAGs from multi‑agent execution logs, incorporates domain priors, detects latent cycles, and outputs uncertainty‑quantified influence structures for downstream modules.

🛠

Key Responsibilities

Design and implement a Bayesian DAG learner that ingests real‑time MAS logs and outputs a causal graph with uncertainty estimates.
Integrate domain knowledge (communication constraints, observability masks) as structural priors and enforce cycle‑consistency via typed‑edge constraints.
Develop online learning pipelines that update the causal graph as new data arrives, ensuring low latency for real‑time blame inference.
Quantify and propagate causal uncertainty to the CGRPA‑Plus and explanation modules, enabling robust blame scores.
Validate the causal discovery engine against synthetic and real‑world datasets, publishing benchmarks and open‑source tooling.

🎯

Required Skills & Experience

Technical Must-Haves

Bayesian network learning (PC, GES, NOTEARS)

Expert

Core to building the causal discovery layer.

Temporal causal inference and dynamic Bayesian networks

Expert

Capturing time‑dependent agent interactions.

Graph neural networks for structure learning

Advanced

Enabling scalable, online inference.

Probabilistic programming (Pyro, Stan, Turing)

Advanced

Implementing Bayesian inference and uncertainty quantification.

Experience Requirements

5+ years leading research in causal inference or probabilistic graphical models.
Track record of publishing in top venues (NeurIPS, ICML, UAI) on causal discovery or dynamic Bayesian networks.
Experience deploying Bayesian models at scale in production or research settings.

Education

PhD in Computer Science, Statistics, or related field with a focus on causal inference or probabilistic modeling.

⭐

Preferred Skills

Experience with online learning algorithms for streaming data.
Knowledge of multi‑agent reinforcement learning and credit assignment challenges.
Familiarity with adversarial robustness evaluation for causal models.

🤝

You Will Thrive Here If...

Thrives in environments where the problem statement is open and evolving.
Comfortable making bold design choices without exhaustive documentation.
Enjoys mentoring junior researchers while owning the entire causal pipeline.

📈

Impact & Growth

12-Month Impact

Within 12 months, deliver a production‑ready causal discovery engine that processes millions of MAS log events per hour, producing a DAG with <5% false‑positive edge rate and <10% uncertainty variance, enabling downstream blame attribution to achieve 30% higher accuracy over baseline methods.

Growth Opportunity

Lead a cross‑disciplinary team that expands CRAN to new domains (autonomous defense, supply‑chain logistics), mentor a research lab, and shape the company’s long‑term causal AI strategy.

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.