Senior Applied Research Engineer - Accelerator Programming Model and Compiler
Core
Designing the next-generation programming model, SDK, and compiler for NVIDIA's Programmable Vision Accelerator (PVA) to support developers and AI coding agents.
Role type
Senior Applied Research Engineer (Compiler & Accelerator Programming)
Builds
PVA programming model, SDK, runtime APIs, and LLVM-based VPU compiler backend
Domain
Accelerated computing, embedded systems, compiler infrastructure, AI coding agents
Required skills
C++ (low-level), LLVM compiler backend development, VLIW/SIMD processor architecture, DSL/IR design, AI coding agent integration, Linux/QNX development, CUDA heterogeneous pipelines
Preferred skills
CUDA integration, OpenCL, MLIR, Halide, TVM, Triton, graph compilers, ISO 26262/IEC 61508 standards, multi-agent workflow orchestration
Responsibilities
Define architecture and feature set of PVA SDK and runtime APIs; Develop and improve LLVM-based VPU compiler backend; Research integration models for PVA workloads in CUDA pipelines; Enable AI agent-driven development via declarative interfaces; Define benchmarks for agent-generated code; Drive technical integrations with customers
Seniority
Senior, hands-on IC
