Agrani Labs is building a high-performance AI accelerator designed for the next generation of intelligent computing. We're looking for an experienced Platform Firmware Architect to join our Platform Firmware Team.
You'll define the system firmware and manageability architecture that makes our AI compute platform secure, observable and serviceable at data-center scale.
What You'll Do
You'll own how our accelerator cards, baseboards and systems are discovered, secured, monitored, updated and recovered in hyperscale and enterprise data centers. This spans the on-device management controller, the Accelerator Management Controller (AMC), the platform root of trust and integration with the host Baseboard Management Controller (BMC). You'll drive conformance to DMTF and OCP specifications and work closely with silicon, SoC firmware, hardware, security, validation, host software and customer teams from architecture definition through production.
Depending on your experience and aptitude, you'll take on one or more of the following:
- Define the end-to-end platform firmware and manageability architecture for AI servers built on Agrani accelerators across multiple product form factors.
- Architect the AMC and on-device management controller firmware, including interfaces with the host BMC, root of trust, SoC management cores and host driver.
- Define in-band and out-of-band management: MCTP over SMBus/I2C, I3C and PCIe VDM; PLDM for monitoring and control, FRU and firmware update; Redfish services conforming to the OCP GPU management profile.
- Architect platform security: hardware root of trust (e.g. Caliptra), secure and measured boot, SPDM authentication and attestation plus NIST SP 800-193 protection, detection and recovery.
- Define fail-safe firmware update and recovery for all platform firmware: PLDM and Redfish update flows, OCP Secure Firmware Recovery, A/B images and anti-rollback.
- Architect telemetry, power, thermal and RAS/debug flows aligned to the OCP GPU & Accelerator requirements, including error logging and crash dumps.
- Lead interoperability with hyperscaler, OEM and ODM BMC stacks (OpenBMC and commercial) and represent Agrani in DMTF and OCP workgroups.
- Set firmware architecture standards, review designs and code, mentor engineers and lead pre- and post-silicon platform bring-up through production.
What We're Looking For
- Bachelor's or Master's degree in Electrical, Electronics or Computer Engineering, Computer Science or a related field
- 10–15 years of hands-on firmware development, with significant recent experience in server platform, BMC or system management firmware and architecture ownership of at least one shipped server or accelerator platform
Good understanding and experience working with
- Server platform architecture: roles of the BMC, AMC and root of trust; management buses (SMBus/I2C, I3C, PCIe VDM, NC-SI); form factors (PCIe CEM, OAM/UBB, DC-SCM); power sequencing, sensors, FRU and event logging.
- DMTF standards: MCTP, PLDM (monitoring and control, FRU, firmware update and RDE), SPDM and Redfish.
- OCP specifications: one or more of GPU & Accelerator Management Interfaces, GPU Firmware Update, GPU & Accelerator RAS, DC-SCM, OAM/UBB, Caliptra and Secure Firmware Recovery.