Research Scientist, Decentralized Training and Inference Verification

Pluralis Research; United States; Remote; USD 120,000 - 180,000; Full time

Job summary

Pluralis Research seeks a Research Scientist to advance decentralized training and inference verification in AI. You will focus on developing verification methods for decentralized training and inference processes, contributing to the integrity and trustworthiness of AI models in distributed environments.

Located in the United States, this role involves defining threat models, designing tests, and building a verification service that operates effectively during inference to ensure robustness and

Qualifications

• Expertise in statistical verification systems for AI.
• Experience with decentralized training and inference verification.
• Ability to define threat models and design tests for reliability.

Responsibilities

• Define threat models for decentralized AI systems.
• Develop verification methods for inference-time reliability.
• Build and operate a verification service for distributed AI models.

Skills

Statistical verification

Distributed AI

Protocol Learning

Job description

Pluralis Research seeks a Research Scientist to advance decentralized training and inference verification in AI. This role is ideal for those with expertise in statistical verification systems and a belief in the potential of Protocol Learning.

In this role, you will focus on developing innovative verification methods for decentralized training and inference processes. Pluralis Research is pioneering Protocol Learning, which involves training large models on consumer-grade devices without central control, posing unique challenges that require robust statistical solutions.

You will be responsible for defining threat models and designing tests to ensure the reliability of decentralized systems. This involves building a verification service that operates effectively during inference, contributing to the integrity and trustworthiness of AI models in a distributed environment.