AI Compiler Optimization Engineer

AI Compiler Optimisation Engineer – Contract (inside IR-35)

Role Overview

We are looking for an experienced AI Compiler Optimization Engineer to enhance the efficiency of AI model inference using advanced compiler techniques. This role involves fine-tuning performance on CPU-based and hybrid CPU/XPU systems, analyzing AI frameworks to uncover optimization opportunities, and leveraging the latest research to drive improvements.

Core Responsibilities

Compiler Optimization for Performance Enhancement:

• Apply compiler techniques such as MLIR and LLVM-based optimizations to improve AI inference performance on CPU and CPU/XPU hybrid architectures.

• Enhance JIT compute graph execution by implementing optimizations like operator fusion and efficient memory management to boost throughput and reduce latency.

• Ideal candidates will have prior experience in LLVM/MLIR development.

AI Framework Profiling & Optimization:

• Analyze AI model execution in frameworks such as TensorFlow, PyTorch, ONNX, and llama.cpp to pinpoint inefficiencies and performance bottlenecks.

• Develop and implement optimizations, including kernel-level enhancements and graph-based improvements.

• Preference for candidates with experience optimizing models across multiple AI frameworks.

Research & Technical Insights:

• Stay informed on emerging trends in AI and compiler technologies by reviewing academic publications and open-source advancements.

• Generate detailed reports that translate cutting-edge research into practical performance enhancements.

• Strong technical writing ability, with prior publications or technical reports, is a plus.

Note: This job description provides an outline of primary duties and expectations. Additional responsibilities may be assigned as required by the company’s evolving needs.

Candidate Profile

Essential Qualifications:

• Strong proficiency in C/C++ with experience in compiler infrastructures such as LLVM or MLIR.

• In-depth understanding of AI model structures and inference execution.

• Hands-on experience with profiling tools (e.g., perf, TensorBoard, VTune) for performance analysis.

• Familiarity with CPU/XPU hardware architectures and related optimization techniques.

• Excellent analytical and problem-solving skills.

Preferred Skills & Experience:

• Degree in Computer Science or a related field (BSc/MSc/MSci).

• Contributions to open-source compiler projects (LLVM/MLIR communities preferred).

• Experience working with heterogeneous computing environments (CPU/GPU/XPU).

• Published research or technical content related to AI and compiler optimization.

• Self-motivated, eager to explore new technologies, and skilled in hands-on problem-solving.

•