I

Vice President, Machine Learning & Systems Compiler Engineering

International Recruiting LLC California, United States

Aug 21
machine-learning Principal (10+ yrs) Full-time United States
Create a free account to apply — email only, no card. You can also save this posting or score it against your profile with AI.

About the role

You will architect and evolve the compiler IR strategy while leading a world-class engineering team to build production-grade toolchains for AI accelerators. The role involves deep collaboration with SoC architecture teams to optimize performance and ensure seamless integration across the hardware-software stack.

What they look for

Compiler Architecture LLVM MLIR RISC-V C++ Python Machine Learning Performance Engineering Graph Optimization Toolchain Development Hardware/Software Co-design Team Leadership SoC Architecture Distributed Systems CI/CD Benchmarking

Requirements

Candidates must have 15+ years of experience in compiler or systems engineering with at least 5 years in a leadership capacity. Expert-level proficiency in C/C++, Python, and hands-on experience with LLVM/MLIR infrastructure is required.

Full description

What You Will Build

Compiler Architecture & Intermediate Representations

● Architect and evolve our client's compiler IR strategy, building on MLIR/LLVM infrastructure to lower high-

level ML graphs into hardware-aware representations for RISC-V and AI accelerator execution

● Design and own custom dialects, optimization passes, and lowering pipelines that map cleanly onto ASHAI's

vector and accelerator ISA extensions

● Set the multi-year technical roadmap for the compiler stack, balancing near-term customer performance

commitments against long-term platform scalability

Graph-Level Optimization & Performance Engineering

● Drive graph transformations, operator fusion, pattern rewriting, quantization, and memory planning strategies

that materially reduce latency and increase throughput on our client's hardware

● Own performance benchmarking of the compiler stack against GPU baselines: tokens per second, cost per

token, and memory efficiency

● Profile, debug, and tune computational kernels, including matrix multiplication, attention, and other core

inference operators, to close the gap between theoretical and achieved performance

Toolchain, SDK & Developer Platform

● Design, build, and maintain a scalable, production-grade compiler toolchain and SDK APIs that customers use

to compile and deploy models on our client's hardware

● Own RISC-V toolchain integration: compiler support, ABI compatibility, and build system tooling for ASHAI-

optimized binaries

● Establish automated testing, benchmarking, and CI/CD infrastructure that keeps the compiler stack stable as it

scalesTeam Building & Technical Leadership

● Recruit, hire, and lead a world-class compiler engineering team from the ground up, setting hiring bar, technical

direction, and career growth paths

● Guide engineers through code review, design review, and mentorship practices that keep the codebase and team

execution at a high standard

● Evaluate and integrate learning-based and agent-assisted approaches into the compilation pipeline and the

team's own development practices where they measurably improve outcomes

Hardware/Software Co-Design & Cross-Functional Collaboration

● Partner directly with the SoC Architecture team to align compiler capabilities with the constraints and

opportunities of our client's custom silicon, informing ISA extension and microarchitecture decisions

● Partner with the VP, System & Applications Engineering on the boundary between the compiler stack and the

runtime, HAL, and developer SDK, keeping API contracts and framework integration paths coherent end to end

● Represent the compiler team's perspective in architecture reviews, ensuring compiler and code-generation

requirements are first-class inputs to silicon decisions

Open-Source Engagement

● Actively participate in, contribute upstream to, and track major open-source compiler ecosystems relevant to

our client's stack, including LLVM, MLIR, OpenXLA/StableHLO, Triton, and TVM

● Represent our client in the broader ML compiler community, building the company's credibility and visibility

among framework and hardware ecosystem partners

What We Are Looking For

Required Experience

● 15+ years in compiler/systems engineering, including 5+ years leading a compiler team

● Expert in modern C/C++ and Python on Linux

● Deep expertise in compiler fundamentals, IRs, and hands-on LLVM/MLIR experience

● Familiarity with major DL frameworks: PyTorch, TensorFlow, or JAX

● Strong foundation in data structures, graph algorithms, and optimization techniques

● Proven ability to build and lead cross-disciplinary engineering teams

Strongly Preferred

● Experience mapping code to GPUs, TPUs, NPUs, or custom AI accelerators

● Background in instruction selection, register allocation, or SIMD/vectorization

● Experience with OpenXLA, StableHLO, TVM, or Triton

● RISC-V depth: privilege levels, RVV extension, LLVM/GCC toolchain, and ABI conventions; ecosystem

contributions a plus

● Track record shipping compiler software to web-scale cloud or embedded environments

● Familiarity with Bazel/CMake and CI/CD pipelines

● Experience taking a new platform's compiler from zero to production-ready

Leadership Qualities

● Operates at both strategic and implementation levels: can set compiler architecture direction and contribute

directly on critical-path components

● Track record of building high-performing compiler or systems teams from a small base, including hiring,

mentoring, and setting technical culture

● Strong communicator across technical and business audiences; able to translate compiler performance work into

customer-facing throughput, cost, and latency terms

● High tolerance for greenfield ambiguity: this role requires defining structure and priorities where none currently

exist

Similar roles