SambaNova is a leader in next-generation AI infrastructure, delivering a full-stack inference platform for customers worldwide. At the core of SambaNova's technology is the RDU (Reconfigurable Dataflow Unit) — a chip built on a dataflow architecture rather than the traditional GPU model. Its decode performance is especially strong for agentic workloads like multi-turn agents, code generation, and long-running applications. RDUs are packaged into SambaRack, rack-scale hardware that lets customers deploy state-of-the-art models with better performance, greater energy efficiency, and faster time to value.

About the team

The RTL design team builds the compute datapath, on-chip network, and memory subsystems inside SambaNova's RDU AI accelerator. Their work spans microarchitecture through tape-out, including RTL implementation, PPA trade-offs, quality gates like lint, CDC, and LEC, and timing closure with physical design.

About the role

SambaNova is looking for a Principal RTL Design Engineer to own the microarchitecture and RTL for our next-generation Reconfigurable Dataflow Unit (RDU), the silicon behind our AI acceleration systems. This is a senior individual contributor role where you'll take a block from microarchitecture specification through RTL, quality closure, and tape-out, owning the power, performance, and area trade-offs along the way. Frequency targets are aggressive, so how you pipeline a datapath and where you spend area to buy timing are essential aspects of the role.

Responsibilities

Own the microarchitecture and RTL for a major block or subsystem of the RDU, from specification through tape-out, including the specs you write for complex blocks in compute, memory, or on-chip network subsystems
Write, integrate, and debug synthesizable SystemVerilog RTL through the full chip implementation and production lifecycle
Own PPA for your area, hitting aggressive frequency targets through datapath pipelining, retiming, and microarchitectural restructuring
Drive timing closure and design quality for your area: synthesis, STA, critical-path iteration, lint, CDC, LEC, and SystemVerilog Assertions capturing design intent
Partner with design verification to define verification intent and coverage for your block, and support debug of functional and gate-level failures
Work with architects and software teams to prototype and evaluate features that improve application-level performance
Drive AI, automation, and new EDA tool methodologies into the design flow
Set the technical bar in design reviews and mentor design engineers

Required Qualifications

B.S. in Electrical Engineering, Computer Engineering, or Computer Science
7+ years of industry experience in ASIC RTL design
Mastery of synthesizable SystemVerilog for high-speed, high-performance ASICs, including SVA to capture design intent
High-frequency design experience, including datapath pipelining, retiming, critical-path restructuring, and other microarchitectural techniques for hitting aggressive frequency targets
Deep foundation in computer architecture and digital logic design, including arbitration, flow control, clock domain crossing, and memory structures
Ownership of a complex block or subsystem from microarchitecture specification through tape-out on at least one production ASIC
Experience converging power, performance, and area against real physical constraints, including floorplan, congestion, and timing feedback from the physical design team
Hands-on experience with synthesis and static timing analysis, and with lint, CDC, and LEC quality flows
Demonstrated technical leadership without formal authority, including setting direction in design reviews, arbitrating microarchitecture decisions, and mentoring engineers
Scripting in Python for design and flow automation

Preferred Qualifications

M.S. or Ph.D. in Electrical Engineering, Computer Engineering, or Computer Science
Depth in one RDU domain: compute and dataflow datapath, on-chip network and interconnect, or memory subsystem
Industry-standard interconnects and protocols including AXI, PCIe, Ethernet, and HBM or DDR