Anthropic seeks a Platform Security Engineer to build and harden the infrastructure that trains and serves frontier AI models. The role owns Dynamic Root of Trust for Measurement (DRTM) across x86 and ARM platforms, implementing attestation services, designing bootloader‑agnostic solutions, and advancing low‑level security features such as confidential computing. The engineer collaborates with firmware, hardware, OS teams, and external vendors, publishes work upstream to the Linux and firmware communities, and contributes to other security initiatives like TDX and SEV.
Platform Security Engineer, DRTM / Secure Launch en Anthropic
Híbrido - San Francisco, CA
Más vacantes en AnthropicSalary
USD 320,000 - 405,000
Requirements
Skills
- Deep hands‑on experience with measured boot and roots of trust: DRTM (Intel TXT, AMD SKINIT, ARM equivalents), SRTM, TPM 1.2/2.0, and the measurement chains built on them
- Strong C and assembly, with deep Linux kernel and early‑boot fundamentals (bootloaders, UEFI, ACPI, SMM)
- A record of landing non‑trivial work upstream in Linux, TianoCore, GRUB, or a comparable community
- Comfort at the hardware/firmware boundary and with the debugging that comes with it: JTAG, serial, platform‑level bring‑up, silicon errata
- Strong technical cross‑functional leadership and direction setting, including with external vendors and OEMs
- Clear written communication: this role produces specifications, design docs, and public technical writing
- Working knowledge of NIST firmware security guidance, particularly SP 800‑193 and 800‑147/155
- 8+ years in systems security, with at least 5 years focused on firmware, bootloader, and OS‑level security
- Existing maintainership or subsystem ownership in Linux, or standing in the TCG, UEFI Forum, or OCP communities
- Confidential computing implementation experience: TDX, SEV‑SNP, ARM CCA, and their attestation flows
- Hardware roots of trust and attestation beyond the TPM: Caliptra, OCP S.A.F.E., SPDM
- Memory‑safe systems code in Rust
- Firmware vulnerability research, reverse engineering, or fuzzing
- Previous work with AI/ML infrastructure security
Responsibilities
- Own adoption and integration of DRTM hardware security features across Anthropic infrastructure, on both x86 and ARM platforms
- Implement attestation services and the additional security and privacy capabilities that DRTM presence enables
- Design and implement a bootloader‑agnostic solution for initiating and relaunching DRTM sessions
- Investigate further hardening of existing DRTM solutions: PPAM to isolate SMM on x86, VMM features to isolate UEFI runtime services, ACPI handling and hardening in DRTM environments
- Interface with vendor and OEM partners on their DRTM solutions: refine requirements jointly and determine which changes are desirable and feasible
- Publish relevant DRTM work upstream and help carry Linux Secure Launch maintainership as tboot is retired
- Act as technical lead in architecture and design, and disseminate the work through technical papers, conference talks, and community engagement
- Assist other Anthropic teams with their own open source efforts and upstream interactions
- Build and harden other platform security features, including confidential computing solutions such as TDX and SEV
- Support custom operating system artifacts, implement device drivers for custom hardware and features, and do firmware diagnosis and enhancement work
- Debug and diagnose kernel and system‑level issues on production hardware
- Take on investigative work in new technical areas and old unsolved ones (for example, the distinct security problems in dTPMs and fTPMs)
Technologies
DRTMIntel TXTAMD SKINITARM equivalentsSRTMTPM 1.2TPM 2.0Linux kernelBootloadersUEFIACPISMMJTAGSerialPlatform‑level bring‑upSilicon errataTDXSEVSEV‑SNPARM CCASPDMRustFirmware vulnerability researchReverse engineeringFuzzingAI/ML infrastructure security
Descubre si tu currículum está listo para esta vacante
Mira cómo nuestra IA puede optimizar tu currículum y aumentar tus chances en este puesto.