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 GPUs connected via the internet. The company has already demonstrated feasibility with Agora, a permissionless run that pretrained an 8B model from scratch on distributed consumer GPUs with no single participant holding full weights.
In a trustless, decentralized network, verification is critical. Workers can cheat by returning tokens from cheaper models, using wrong sampling parameters, dropping updates, flooding the system, submitting trivial work for rewards, poisoning updates with backdoors, extracting private data from gradients, or inflating reported contributions. This role owns the verification layer that ensures training and inference integrity across the entire network.
You will:
- Own and evolve the threat model, enumerating what malicious or careless workers can do across pre-training, post-training, and inference (training disruption, free-riding, model poisoning, data extraction, reward manipulation).
- Design and calibrate statistical verification methods with stated error rates, tuning them rigorously across heterogeneous hardware (different GPUs, Macs) to keep false positives and false negatives controlled.
- Build and operate the verification service in the inference path, taking ownership of its real-world performance and mistakes.
This is a foundational research and systems role at the intersection of distributed systems, cryptography, statistics, and adversarial ML. You will write some of the first published answers to problems that mostly have no existing solutions.
REQUIREMENTS:
- Shipped or published verification systems: You have built a calibrated statistical decision system with stated error rates and operated it in production. Publications in inference and training verification, fraud detection, anti-cheat systems, or experimentation platforms all count equally with academic papers.
- Deep statistical expertise: Strong foundation in statistics and probability; ability to design experiments, calibrate decision thresholds, and defend claimed error rates.
- Technical breadth in verification: Knowledge of the solution space for verifying untrusted compute, including statistical testing, re-execution, cryptographic proofs, and trusted hardware; ability to argue what fits a permissionless network.
- Mission alignment: Belief that Protocol Learning is a viable third path for collective, trustless, and sovereign AI.
NICE TO HAVE:
- Familiarity with large-scale pre-training and RL post-training.
- Experience with decentralized ML security and adversarial threat models (poisoning, Sybil attacks, collusion, replay).
- Background at proprietary, open-weight, or open-source AI labs.
The company is backed by Union Square Ventures and other tier-1 investors. Team is distributed globally with main hubs in Australia and North America; you must be comfortable working across timezones. Professional-level English proficiency (written and spoken) required.
