Principal Engineer, Applied Research - Accelerator Programming Model and Compiler

Details

As the world’s leading accelerated computing company, NVIDIA is building a team to define the next-generation programming model for the Programmable Vision Accelerator (PVA). The PVA is a power-efficient, deterministic VLIW/SIMD processor used across NVIDIA DRIVE, Jetson, and IGX platforms. This role sits at the boundary between research and production systems software and focuses on programming models, compiler/runtime interfaces, and integrations with AI coding agents.

Responsibilities

• Define the next-generation PVA programming model for developers and AI coding agents to make it easier to build optimized algorithms for PVA.
• Enable AI agent–driven PVA development by abstracting hardware-specific details into agent-accessible, declarative interfaces; improve compile time, emulator speed, and diagnostic tooling.
• Provide technical leadership for the architecture and feature set of the PVA SDK, runtime APIs, and programming model.
• Define benchmarks and evaluations for agent-generated PVA code and use results to improve compiler optimizations and diagnostics.
• Develop and actively improve the LLVM-based VPU compiler backend targeting a VLIW/SIMD architecture.
• Research and define efficient integration models for PVA workloads in CUDA-based heterogeneous pipelines, including execution, memory movement, and synchronization.
• Drive deep technical integrations with internal and external customers to improve adoption and shape the PVA runtime API for real-world workloads.

Requirements

• BS, MS, or PhD in Computer Science, Computer Engineering, Electrical Engineering, or related field, or equivalent experience.
• 15+ years of experience building high-performance, low-level systems software, accelerator software, embedded software, or compiler/toolchain infrastructure.
• Experience building higher-level programming abstractions, domain-specific languages (DSLs) or compiler IRs that abstract hardware complexity while preserving performance.
• Experience with compiler, debugger, linker, or toolchain development, particularly using LLVM.
• Experience developing code with agents — Claude Code, Cursor, OpenAI and inference SDKs — and integrating compilers/developer tools with AI coding agents or agentic development harnesses.
• Experience programming SIMD/VLIW processors.
• Excellent software development skills in C++, including low-level debugging and performance profiling.
• Experience with Linux or QNX development environments.
• Strong communication and collaboration skills.

Ways to Stand Out

• Experience with CUDA, especially integrating accelerators into a CUDA-based heterogeneous compute pipeline.
• Background with MLIR, Halide, TVM, Triton, graph compilers, image-processing DSLs, or other declarative/compiler-based programming systems.
• Familiarity with ISO 26262 and IEC 61508 or equivalent quality/safety standards.
• Experience building agent harnesses and orchestrating long-running, stateful, multi-agent workflows.

Compensation & Benefits

• Base salary range: 272,000 USD - 431,250 USD (final base salary will be determined based on location, experience, and pay of employees in similar positions).
• Eligible for equity and company benefits (link to NVIDIA benefits on the original posting).

Additional Information

• Applications accepted at least until July 20, 2026.
• This posting is for an existing vacancy.
• NVIDIA uses AI tools in its recruiting processes.
• NVIDIA is an equal opportunity employer and committed to fostering an inclusive work environment.