Build sandboxed environments that wrap real Niural workflows, stateful across episodes and seeded for reproducibility.
Design programmatic verifiers from known ground truth, scoring trajectories to prevent reward hacking.
Train agents via RL loops, curriculum schedules, and fine-tuning, then publish negative results internally.
Niural is a global Payroll, Employer of Record (EOR), Agent of Record (AOR), and Contractor Management platform that empowers businesses in the digital economy. We are a team building foundational internet infrastructure with a focus on speed, ownership, and ambition.
Build the environments and verifiers our models train against
Own the synthetic data pipeline, from generation through quality gates
Ship models into production and keep improving them on real usage
LiveKit builds the infrastructure layer for the voice-driven era of computing, enabling developers to build, deploy, and scale voice AI applications. Founded in 2021, they power voice AI for major companies and have a small, senior team that values craft and creativity.
Build new RL environments targeting different agentic capabilities and industry areas.
Train and evaluate agents in those environments, improving both agents and environments.
Work across modeling and product to identify performance gaps and automate measurement.
Cohere is a security-first enterprise AI company building cutting-edge foundation models and end-to-end products. It is a global team of researchers, engineers, and designers headquartered in Toronto with offices worldwide.
Build systems that estimate causal response, choose actions, and generate useful information while updating policies and deploying within guardrails.
Work across causal modeling, uncertainty estimation, contextual bandits, policy learning, and production ML infrastructure.
Success is measured by measurable economic lift in controlled experiments, safe automation, and systems that improve by operating the business.
CSC Generation builds closed-loop decision systems that use machine learning to operate consumer businesses more intelligently. The company focuses on areas such as pricing, inventory, purchasing, promotions, marketing, and assortment.
Design and implement tool-based environments for knowledge work simulations.
Write clean and modular code to support reinforcement learning training pipelines.
Troubleshoot and refine environment mechanics based on testing feedback.
Terac is building the world's largest pool of vetted human experts for AI. They are a platform for researchers and AI labs to recruit, screen, and pay study participants across industries and skill sets.
Co-direct the research program in partnership with the founder to shape scientific vision and technical execution.
Develop RL feedback loops translating high-throughput biological measurements into reward signals for biological language models.
Perform hands-on engineering including architecting model training infrastructure and building, running, and debugging models.
Astera is a private foundation steering science and technology toward an abundant future by supporting projects as high-velocity startups. It operates unlike traditional non-profits, focusing on ambitious goals with strong technical talent and significant resources.