SlipstreamJobs tracks this role from the company's public career site. Apply directly on the employer's site.
Salary: USD 150,000 - 300,000 / annual
Prime Intellect is building an open superintelligence stack—infrastructure that frontier AI labs build internally, now available to every ambitious AI team. The platform unifies compute, environments, evaluations, secure sandboxes, high-performance training, and deployment for post-training at frontier scale. The company has raised $150M from Founders Fund, Radical Ventures, NVIDIA, and notable operators including Andrej Karpathy and leaders from Perplexity, Harvey, Cognition, OpenAI, and others.
In this Applied Research role focused on chip design, you will build agents, environments, evaluations, and post-training systems for real semiconductor design and verification workflows. Chip design is a natural fit for agentic reinforcement learning: it involves long-horizon work, specialized tools, expensive mistakes, hard constraints, and objective feedback from simulation, verification, timing, power, and area analysis.
Your work will be grounded in external design workflows with access to domain experts and toolchains. You may span RTL generation and debugging, formal verification, simulation, physical design, optimization, and orchestration across EDA tools.
Applied Research at Prime Intellect turns workflows into tasks, environments, evaluations, and reward signals, runs post-training experiments against real baselines, and deploys what works. Every Applied Research role is externally facing—most researchers work directly with customers, users, or research partners while maintaining a substantive research agenda.
You will own: interactive environments where agents use EDA tools to modify designs, debug failures, and plan multi-step work; verifiers and rewards grounded in tests, simulation, formal methods, design rules, and engineering outcomes; experiments on credit assignment when feedback arrives late and multiple tools contribute; and evaluations for correctness, recovery, and generalization with reusable methods for other simulator-grounded engineering domains.
Representative early problems include building an interactive verification environment where an agent reads a specification, writes or repairs RTL, invokes simulation or formal tools, interprets failures, and iterates toward a verified result; designing rewards and evaluations that combine functional correctness with tool efficiency, robustness, and performance/power/area constraints; and training long-horizon agents to recover from design and tool failures, preserve useful state, and generalize across blocks, designs, and constraint sets.
REQUIREMENTS:
- Depth in RL, agents, or post-training systems and the ability to become fluent in semiconductor workflows
- Prior work in design automation, compilers, formal methods, or ML for systems is useful but not required
- Hands-on experience building and debugging model-training, post-training, agent, evaluation, or closely adjacent experimental loops
- Strong experimental judgment: meaningful baselines, careful trace inspection, clear failure attribution, and willingness to change direction when evidence changes
- Strong Python and ability to build artifacts that other researchers and engineers can run, inspect, and extend
- Comfort owning ambiguous problems and keeping partners and teammates current as you resolve them
- Evidence of exceptional work valued: publications, open-source systems, benchmarks, and shipped products
