Senior Principal Hardware Emulation Engineer

Senior Principal Hardware Emulation Engineer

San Jose, California, United States

The era of pervasive AI has arrived. In this era, organizations will use generative AI to unlock hidden value in their data, accelerate processes, reduce costs, drive efficiency and innovation to fundamentally transform their businesses and operations at scale.

Samba Nova Suite™ is the first full-stack, generative AI platform, from chip to model, optimized for enterprise and government organizations. Powered by the intelligent SN40L chip, the Samba Nova Suite is a fully integrated platform, delivered on-premises or in the cloud, combined with state-of-the-art open-source models that can be easily and securely fine-tuned using customer data for greater accuracy. Once adapted with customer data, customers retain model ownership in perpetuity, so they can turn generative AI into one of their most valuable assets.

The emulation team builds the pre-silicon model of Samba Nova's RDU that the software, architecture, and verification teams depend on.

Samba Nova is looking for a Senior Principal Engineer to build and own the hardware emulation models for our next-generation Reconfigurable Dataflow Unit — the silicon behind our AI workload acceleration systems.

Emulation is a development platform here, not a checkbox. Our software teams bring up drivers, runtime, and compiler against the emulation model months before silicon. Our architects measure performance on it. Our verification team runs on it to get through scenarios that simulation cannot reach in useful time. When the model is fast, accurate, and available, three organizations move faster; when it is not, they wait.

Building that model is the core of the job. You will work shoulder to shoulder with the RTL design team to keep the design emulation-compliant, decide where a vendor model will do and where you need to write a custom one, and own the automation that keeps the platform running tests continuously rather than by hand. This is a senior individual-contributor role, and much of its leverage comes through other teams: you will be the person design, verification, and software engineers consult on what emulation can and should do, and the one who holds our emulator vendors to account.

Responsibilities

Build the Models

• Develop the emulation model for each new RDU generation: high-performance transactors, behavioral models, and memory models in System Verilog and C/C++
• Partner with the RTL design team to keep the design emulation-compliant: review RTL and IP for constructs that will not map cleanly to the emulator, and close the gaps by substituting vendor emulation models or hand-coding custom models where no vendor model exists
• Own model fidelity: make sure substituted and custom models track design intent closely enough that software, performance, and verification results can be trusted
• Influence design decisions upstream so emulation-readiness is designed in rather than worked around

Own the Platform, Flows, and Automation

• Own the emulation environment architecture: build and runtime flows, partitioning, compile strategy, clocking, and memory configuration for each chip generation
• Build and own the workflows that automate emulation testing on the hardware: job scheduling, regression launch, failure triage, and results reporting, so the platform runs continuously without manual babysitting
• Own throughput and capacity: profile and remove bottlenecks in models, transactors, and testbench collateral so turnaround stays fast as the design grows
• Own the emulator vendor relationship: drive blocking tool issues to resolution, hold the vendor accountable to our program schedule, and advise on platform selection and capacity investment

Serve the Three Primary Use Cases

• Software and hardware co-development: deliver and sustain the platform software teams depend on for pre-silicon bring-up of drivers, runtime, and compiler, and debug hardware/software integration failures alongside them
• Performance verification: enable workload bring-up, measurement methodology, and reproducible results, and work with architecture to correlate emulated performance against…