Senior Distributed Inference & Protocol Learning Engineer

Pluralis Research is pioneering Protocol Learning, a decentralized approach to training and serving large models on consumer-grade devices connected via the internet. Despite initial skepticism, they’ve achieved groundbreaking progress, including Agora, a permissionless run that pretrained an 8B model from scratch using consumer GPUs without any single participant holding full weights. Explore their mission in full via A Third Path: Protocol Learning.

The inference pipeline currently generates rollouts for reinforcement learning (RL) training and will serve their models post-training. This system operates in a permissionless, trustless environment, introducing unique challenges: hardware consists of Macs and consumer GPUs owned by strangers, the network is the public internet, nodes dynamically join/leave mid-run, and weights evolve during training. Your role is to design and build systems ensuring this pipeline remains fast and reliable under these conditions.

Key Responsibilities

• Own the inference stack end-to-end: Develop and manage pipeline-parallel execution, placement/routing, transport protocols, the serving engine, and failure handling. Drive direction and execution.

• Invent novel algorithms: Create methods for fast inference on consumer hardware over the public internet, validate them, and deploy in production.

• Serve training and users: Ensure the rollout pipeline is fast and reliable for RL training today, and transform it into the serving layer for trained models.

Requirements

• Shipped serving systems: Hands-on experience building large-scale inference systems or serving-engine internals.

• Research background: Publications (papers/blogposts) in distributed inference, LLM serving systems, pipeline parallelism over slow networks, or decentralized training. Unpublished work is also considered.

• Low-bandwidth networking: Experience with systems operating in low-bandwidth, high-latency environments like the public internet.

• Mission alignment: Belief in Protocol Learning as a viable, trustless and sovereign path for collective AI.

Nice to Have

• Familiarity with RL post-training workflows.

• Experience with Apple silicon or MLX.

• Knowledge of P2P networking and NAT traversal.

• Background in proprietary, open-weight, or open-source AI labs.

Compensation & Benefits

• Equity-Heavy Package: Significant ownership for key technical contributors, alongside a high base salary.

• Remote-First Culture: Flexible work environment with global team distribution.

• Visa Sponsorship: Optional full visa sponsorship and relocation support to Australia or the US.

• Open Problems: Opportunity to solve unanswered challenges in training/serving frontier models on uncontrolled hardware/networks.

Additional Notes

• Teams are distributed globally, primarily in Australia and North America. Timezone flexibility is required.

• Professional-level English proficiency (written/spoken) is mandatory.

• No agency support is currently accepted.