Research Scientist - Geo-Distributed Inference
AI
Pluralis Research is seeking a Research Scientist to own and build the inference stack for Protocol Learning, a decentralized approach to training and serving large models on consumer hardware over the public internet. The role involves designing novel algorithms for pipeline-parallel execution, placement, routing, transport, and failure handling in a permissionless, trustless setting. Candidates should have hands-on experience shipping serving systems, research publications in distributed inference or related fields, and a strong alignment with the mission of collective, trustless AI.
What you'll do
- Own the inference stack end-to-end: pipeline-parallel execution, placement and routing, transport, serving engine, and failure handling
- Invent algorithms for fast inference on consumer hardware over the public internet
- Serve training and users: keep rollout pipeline fast and reliable for RL training and turn it into serving layer for models
What you need
Must-have
Distributed InferenceLLM Serving SystemsPipeline ParallelismLow-Bandwidth NetworkingDecentralized TrainingReinforcement LearningFailure HandlingPlacement and RoutingTransport Protocols
Nice-to-have
RL Post-TrainingApple SiliconMLXP2P NetworkingNAT Traversal
Soft skills
Mission AlignmentComfortable working across timezonesProfessional-level English proficiency
Research ScientistDistributed InferenceProtocol LearningDecentralized AILLM ServingReinforcement LearningRemoteFull-timeAustraliaTechnology