SlipstreamJobs tracks this role from the company's public career site. Apply directly on the employer's site.
Pluralis Research is building Protocol Learning: a system for training and serving large language models in a fully decentralized way on consumer-grade devices connected via the internet. The company has already demonstrated feasibility with Agora, a permissionless run that pretrained an 8B model from scratch on consumer GPUs distributed across the internet, with no single participant holding the full weights.
The inference pipeline currently generates rollouts for reinforcement learning training and will serve models once trained. It operates in a permissionless, trustless setting where hardware consists of Macs and consumer GPUs owned by strangers, the network is the public internet, nodes join and leave mid-run, and model weights change during training. This creates unique systems challenges that require novel solutions.
As Research Scientist for Geo-Distributed Inference, you will own the inference stack end-to-end: pipeline-parallel execution, placement and routing, transport layer, serving engine, and failure handling. You will set technical direction and drive implementation. You'll invent new algorithms to make inference fast on consumer hardware over the public internet—methods that don't yet exist in published literature. You'll keep the rollout pipeline fast and reliable for RL training today, and evolve it into the serving layer for production models.
The role is deeply technical and hands-on. You'll work across timezones with teams distributed globally, primarily in Australia and North America. The company is backed by Union Square Ventures and other tier-1 investors, and the team consists of world-class ML researchers. Pluralis is ideologically committed to implementing a decentralized protocol for intelligence.
REQUIREMENTS:
- Shipped serving systems: hands-on experience building serving-engine internals or large-scale inference systems
- Research ability: publications (papers or blog posts) in distributed inference, LLM serving systems, pipeline parallelism over slow networks, or decentralized training; unpublished work you can walk through is also valued
- Low-bandwidth networking: experience with systems running in low-bandwidth, high-latency settings like the public internet
- Mission alignment: belief that Protocol Learning is a viable third path for collective, trustless, and sovereign AI
- Professional-level English proficiency (written and spoken)
NICE TO HAVE:
- Familiarity with RL post-training
- Exposure to Apple silicon or MLX
- Experience with P2P networking and NAT traversal
- Experience at proprietary, open-weight, or open-source AI labs
