[Hiring] Research Scientist - Decentralized Training and Inference Verification @Pluralis Research

Role Description

As a key member of Pluralis Research, your primary role is to develop efficient algorithms and systems that verify the work within our decentralized Protocol Learning framework.

• Own the threat model: Enumerate potential malicious or careless actions by workers across pre-training, post-training, and inference, including:
  • Training disruption and denial-of-service
  • Free-riding
  • Model poisoning and backdoors
  • Data extraction from gradients and activations
  • Reputation and reward manipulation
• Design and calibrate the tests: Build statistical verification methods with stated error rates, tuning them with rigorous benchmarks to control false positives and false negatives across heterogeneous hardware.
• Ship the verifier: Build and run the verification service in the inference path, accepting responsibility for its mistakes.

Qualifications

• Verification systems, shipped or published: Experience in building calibrated statistical decision systems with stated error rates. Publications in inference and training verification are valued.
• Statistical depth: Deep expertise in statistics and probability, with the ability to design experiments and defend claimed error rates.
• Technical background: Knowledge of verifying untrusted compute solutions, including statistical testing, re-execution, cryptographic proofs, and trusted hardware.
• Mission alignment: Belief in Protocol Learning as a viable third path for collective, trustless, and sovereign AI.

Requirements

• Familiarity with large scale pre-training and RL post-training.
• Familiarity with decentralized ML security and adversarial threat models, such as poisoning, Sybil, collusion, and replay.
• Experience at proprietary, open-weight, and open-source AI labs.

Benefits

• Equity-Heavy Package: Significant ownership for key technical contributors in addition to a high base salary.
• Remote-First Culture: Flexible work environment with team members distributed globally.
• Visa Sponsorship: Optional full visa sponsorship and relocation support to either Australia or the US.
• Open Problems: Opportunity to write some of the first answers to training and serving frontier models on hardware and networks you don't control.