Build and post-train agents for scientific research, including long-horizon environments, evaluations, verifiers, and reward signals based on research partners’ workflows and data. Develop and rigorously evaluate RL and multi-agent approaches, maintain reproducible experiments, and release reusable research artifacts.
Build agent environments, verifiers, rewards, and evaluations for semiconductor design and verification workflows using EDA tools. Run post-training experiments, assess results against baselines, and develop systems that improve agents’ correctness, recovery, and generalization.
Own the development of open-source reinforcement learning, environment, evaluation, and agent systems, and build researcher-facing features for Prime's training platform. Work with external researchers and contributors to create stable interfaces, technical resources, and a coherent path from task design through evaluation, training, and deployment.
Lead applied reinforcement learning programs from customer discovery through environment and evaluation design, post-training experiments, deployment, and measurement of practical outcomes. Identify which components drive results and turn reusable findings into shared research, open-source systems, and product capabilities.
Build and evaluate continual-learning systems that convert agent trajectories, corrections, and failures into training tasks and model updates, while measuring retention, transfer, forgetting, and general capabilities. Develop reproducible training and evaluation pipelines, establish deployment promotion and rollback gates, and conduct research with external collaborators.
Build simulation and real-world environments, evaluations, agent systems, and post-training methods for long-horizon robotics tasks. Train and assess embodied agents using reinforcement learning and other methods, and improve their ability to adapt, recover, and operate safely under real-world constraints.
Build and deploy post-training and agent systems for public-sector and regulated-industry partners, including private improvement loops, permissioned agents, and high-assurance evaluations. Run experiments within partner data and infrastructure constraints, assess results against meaningful baselines, and turn reusable findings into research, open-source systems, and product capabilities.
Build multimodal computer-use agents and post-training and evaluation systems for browser, desktop, and cross-application workflows. Create workflow-based environments, rewards, and experiments, then assess and deploy improvements while ensuring reliable recovery, completion, and safe boundary enforcement.
Build quantitative research environments, agents, evaluations, and post-training methods for systematic decision-making in finance. Develop realistic simulations and production feedback loops, and assess whether results are robust across data, market regimes, risk constraints, and infrastructure.
Develop and test reinforcement learning and post-training methods for tool-using, multi-agent, memory-augmented, and long-horizon systems. Build reusable environments, harnesses, rewards, and evaluations, and turn validated research into open-source systems, papers, and product capabilities.
Design and implement challenging coding and long-horizon agent tasks, executable environments, evaluation systems, and scoring logic for training and assessing Prime Intellect models. Build and curate RL data pipelines, investigate model and harness failures, and collaborate with researchers to test whether task and data changes improve performance on held-out tasks.
Build and optimize inference systems for Prime Intellect’s reinforcement-learning research, including support for new model architectures, long-horizon agent rollouts, and multi-node mixture-of-experts inference. Improve throughput, weight-update efficiency, and numerical reliability while collaborating with researchers on inference-efficient model architectures.
Find and process web, code, and technical data into high-quality, training-ready corpora, with scalable pipelines and clear provenance and versioning. Collaborate with researchers to design data mixtures and run controlled experiments that measure how data choices affect pretraining, reasoning, coding, and long-context model performance.
Develop pretraining code and take model experiments through full-scale training runs, evaluating architecture changes and improving throughput and resilience. Investigate training dynamics, run reproducible ablations, and collaborate on data mixtures, midtraining, and long-context experiments.
Develop and evaluate reinforcement learning algorithms and training systems for open models, including methods for reasoning, coding, and long-horizon tasks. Design scaling experiments, improve the prime-rl framework, and diagnose stability, performance, and numerical issues across training and rollout systems.
You will own strategic customer acquisition and expansion for GPU infrastructure and the Model Factory platform by building relationships with technical executives. You will negotiate multi-year, high-value agreements and partner with solutions architects to translate customer roadmaps into committed capacity.
You will lead technical discovery and design customer architectures on Prime infrastructure to ensure technical wins. You are responsible for owning workload benchmarks, proofs of concept, and creating repeatable playbooks for platform adoption.
Build and optimize distributed training infrastructure for pre-training and large-scale RL workloads using the prime-rl framework. Design low-level performance optimizations and improve end-to-end efficiency across compute, memory, and networking layers.
Lead the growth organization by overseeing sales, marketing, partnerships, and customer success to connect technology to the market. Define GTM strategies for RL infrastructure and close large-scale compute and post-training contracts.
Build the open superintelligence stack and infrastructure for frontier AI labs. Develop open-source models and a full-stack platform for post-training, agent workflows, and autonomous research.
Build and maintain the company's data warehouse and pipelines to create a live, accurate picture of compute supply and demand. Develop data models and dashboards to enable cross-functional teams to track utilization and capacity planning.
Manage a portfolio of enterprise customers to ensure successful adoption, retention, and expansion of AI infrastructure services. Act as a technical partner to optimize training and inference workloads while translating customer needs into product feedback.
Design and operate Kubernetes-based training and inference orchestration across multi-cloud GPU fleets. Build developer-facing surfaces for job submission, monitoring, and model management using a modern full-stack.
You will own the analytical foundation for global compute markets, including pricing supply, modeling economics, and evaluating hardware generations. You will partner with leadership to drive strategic capital allocation and manage commercial diligence with cloud providers.
You will own the end-to-end compute strategy, including sourcing, economics, and contracting for GPU capacity to power the company's AI infrastructure. You will also partner with research and engineering teams to align supply with training roadmaps and technical requirements.
You will define the brand narrative and lead developer relations to drive adoption of Prime Intellect's open superintelligence infrastructure. This involves managing content engines, executing product launches, and building a high-performing marketing team.
