Microsoft looks to be on a hiring spree for engineers and researchers to develop optical and photonic interconnects to reshape AI networking.
Job listings seen by SDxCentral seek staff to help pioneer hardware and system technologies from scale-up networking, co-packaged optics (CPO), and connectivity chip architectures.
The senior-most role on offer is a researcher for “AI computer architectures,” with prospective staff to define Microsoft’s research roadmap and long-term agenda for novel optical interconnects capable of “breaking through the bandwidth, latency, and energy walls.”
Beyond networking, the successful candidate would also explore memory disaggregation and off techniques, specifically addressing the infrastructure requirements for agentic AI workloads. Microsoft wants candidates with experience in scale-up networking across both graphic processing units (GPUs) and general processing units (XPUs), as well as memory systems for AI inference and training infrastructure.
Among the more engineering-focused roles on offer are opto-electronics engineers tasked with designing optical and photonic systems spanning photonics, integrated circuit design, and signal processing. A device engineering role would build hardware that included microLEDs, lasers, and photodetectors for high-speed optical communication applications.
Rounding out the team is a principal optical packaging engineer role, tasked with bridging photonic devices, electronic integrated circuits, and system-level integration through scalable, manufacturable packaging solutions. The hyperscaler wants staff with hands-on experience working with optical modules, CPO packaging designs, or fiber attach/optical coupling.
“You will join a team at the frontier of AI system design, where novel photonic hardware meets the rapidly evolving demands of large-scale AI workloads,” the senior role job description reads. “Your ideas won't stay on paper – you will work alongside world-class researchers and industry partners to translate architectural vision into working silicon and system prototypes that define the next generation of AI infrastructure.”
The roles are all being offered as part of the hyperscaler’s Future AI Infrastructure (FAI) team in Cambridge, England. The site is already home to Microsoft’s existing photonic projects, including its analog optical computer, which uses microLED and sensors from smartphone cameras to power AI and logic computations exponentially faster than conventional computers.
Microsoft is also exploring using microLED for data center interconnects. Its Mosaic prototype, unveiled last September, uses commercially available microLEDs combined with imaging fiber to support high-speed network connections across further reaches than copper while using far less power.
Research efforts in Cambridge will eventually come to production, with Microsoft confirming in March it plans to work with MediaTek to make its microLED interconnects available to partners by late 2027.