Jobs at Thinking Machines Lab

Coordinate high-volume interview scheduling across time zones, manage logistics (travel, expenses, rooms, technical support), own candidate communications, scale and train interviewer pools, use scheduling tools to automate processes, collect and audit interview feedback, maintain recruiting documentation and data integrity, and provide ad-hoc operational support for recruiting initiatives.

Build, operate, and maintain research infrastructure (evaluation frameworks, RL training systems, experiment tracking, visualization). Develop scalable distributed pipelines, ensure reproducibility and observability, and partner with researchers and infrastructure teams to accelerate ML research and tooling adoption.

The Executive Business Partner will manage calendars, coordinate travel, support recruiting, and track projects for technical leaders in a fast-paced startup environment.

Build and operate ML research infrastructure: evaluation frameworks, RL training systems, experiment tracking, distributed evaluation pipelines, observability, and reproducibility tooling. Partner with researchers to identify bottlenecks, drive adoption, and integrate tools across infrastructure, data, and product stacks.

Diagnose and remediate hardware, firmware, and OS issues across large GPU clusters. Own drivers, kernel interfaces, diagnostics, and firmware lifecycle. Automate reliability monitoring, analyze error rates, engage vendors and manage RMAs, and write postmortems to reduce failures and improve fleet reliability for large-scale AI experiments.

Serve as the primary onsite IT engineer for the New York office, handling employee support, onboarding/offboarding, Okta identity management, device fleet management (macOS/MDM), office networking/firewalls, A/V and physical security, Terraform/Git-based workflows for SaaS, vendor coordination, and cross-office collaboration with SF IT and Security teams.

Design, build, and operate macOS endpoint infrastructure and MDM as a production service. Lead MDM migration, manage Santa/Rudolph, implement zero-touch provisioning, continuous posture evaluation, automated patching, software distribution, fleet telemetry, and remediation. Drive automation, observability, incident response, advanced macOS troubleshooting, and technical leadership across IT, Security, Identity, and Infrastructure.

Own and debug low-level GPU network fabric (RDMA/RoCE, NVLink/NVSwitch) for large-scale training/inference. Build host instrumentation, triage cross-cloud networking issues, and drive escalations with cloud providers to maintain interconnect reliability at scale.

Build and manage pipelines for research, engineering, and infrastructure talent. Partner with recruiters and hiring managers, craft personalized outreach to passive candidates, track sourcing metrics, represent company mission, and help develop sourcing tools and processes at a fast-scaling startup.

The Site Reliability Engineer will drive reliability for the Tinker platform, focusing on incident response, monitoring, and ensuring system resilience while collaborating across teams.

As a Research Product Manager, drive complex research products, translate technical ideas into plans, collaborate across disciplines, and maintain progress communication.

The Software Engineer, Data Infrastructure will design scalable systems for data processing and orchestration, collaborating with teams to enhance data quality and accelerate research.

The role involves building tools for AI-assisted software development, optimizing workflow, and maintaining secure development environments.

The Forward Deployed Engineer will support customers by diagnosing issues, improving products, building tools, and informing Tinker's roadmap.

The role focuses on designing and implementing data curation techniques for AI pre-training datasets, ensuring quality, collaboration, and continuous improvement of data processing systems.

The Research Engineer will design, optimize, and scale systems for large AI models, enhancing performance, reliability, and efficiency in model inference and deployment.

Design and optimize large-scale distributed training infrastructure, focusing on numerical methods to enhance efficiency and stability in AI model training.

The role involves designing and optimizing distributed training systems for large models to enhance research efficiency and productivity in AI.

Design, build, and operate GPU supercomputing environments, automate large GPU clusters, and collaborate with researchers to optimize performance and resources.

The Software Engineer at Tinker will develop the platform systems for billing, permissions, and data management. Responsibilities include designing authorization layers and managing end-to-end billing infrastructure, collaborating with teams, and ensuring compliance with enterprise requirements.