Labs want low‑level systems talent
Social signals show OpenAI and Anthropic actively hiring firmware, kernel, bare‑metal ops and fleet management engineers—roles that support scaling compute and runtime reliability rather than model research alone. That push confirms elite labs are investing in foundational infra talent to run large experiments and production fleets. (x.com) (x.com)
OpenAI’s live listing for “Software Engineer, Fleet Management” places the role in San Francisco, requires hybrid on-site work three days per week, and explicitly tasks engineers with integrating low-level hardware metrics into cluster scheduling and orchestration. (openai.com) OpenAI is also advertising firmware- and NOS-focused roles that name SONiC, switch ASIC SAI/SDKs and kernel-level work as core responsibilities for bootstrapping and scaling switching layers of its compute fabric. (simplyhired.com) Anthropic’s TPU Kernel Engineer posting lists San Francisco, New York City and Seattle as locations, requires at least a bachelor’s or equivalent experience, and shows an advertised annual compensation band of $280,000–$850,000 for the role. (job-boards.greenhouse.io) Anthropic’s compute/infra openings include a Platform Hardware Security role that explicitly mentions firmware, bootloader and attestation system design to secure bare‑metal infrastructure. (itspronounceddata.getro.com) Anthropic’s public jobs page shows dozens of simultaneous openings—61 roles in AI Research & Engineering and 11 in Compute—indicating parallel hiring across both research and low‑level infrastructure teams. (anthropic.com) Research scientist job descriptions at DeepMind and OpenAI continue to list a PhD (or equivalent research track record), publications at top conferences, and strong mathematical and software-engineering skills as baseline expectations for senior research hires. (job-boards.greenhouse.io) Recruiting guides and market listings show elite-research hires are still evaluated on conference publications and theoretical fluency while systems roles prioritize kernel-level optimization, accelerator/kernel experience, and large-scale systems debugging—different objective signal sets for hiring committees. (datainterview.com)
