06 Aug
|
Recognized
|
Bengaluru
06 Aug
Recognized
Bengaluru
About the Company
We are hiring on behalf of our client, a global technology leader known for developing cutting -edge semiconductor, AI, graphics, and compute solutions powering next -generation mobile, automotive, IoT, and intelligent computing platforms. The company is at the forefront of AI acceleration, GPU architecture, and high -performance computing innovations.
Role Overview
We are seeking a GPU Compute & MLIR Compiler Engineer with strong expertise in GPU programming, MLIR compiler development, and AI workload optimization. The role involves designing and optimizing GPU compute kernels, developing MLIR -based compilation pipelines, and driving performance improvements for AI/ML applications across modern GPU architectures.
Key Responsibilities
- Develop and optimize GPU compute kernels for AI/ML workloads using OpenCL and Vulkan Compute.
- Design, build, and extend MLIR dialects and compiler infrastructure across multiple abstraction levels.
- Develop compiler passes including tiling, fusion, vectorization, bufferization, and lowering pipelines.
- Optimize AI model execution through compiler -level and runtime -level performance enhancements.
- Perform GPU profiling, bottleneck analysis, and performance tuning using hardware counters and profiling tools.
- Build and maintain GPU runtime components, including memory management, resource scheduling, and execution pipelines.
- Develop efficient code generation pipelines from MLIR to GPU backends.
- Collaborate with ML, runtime, framework, and hardware teams to optimize CV and LLM workloads.
- Develop high -performance compiler and runtime components using modern C/C++.
Required Skills
Compiler & MLIR
- Strong experience with MLIR framework.
- Hands -on experience creating custom dialects, compiler passes, and lowering pipelines.
- Knowledge of:
- TOSA, StableHLO, ONNX -MLIR
- Linalg, Tensor, Vector Dialects
- MemRef, SCF, GPU, LLVM Dialects
- Graph optimization and code generation techniques
GPU Programming
- Robust expertise in OpenCL kernel development and optimization.
- Experience with Vulkan Compute programming and runtime internals.
- Understanding of GPU architecture, memory hierarchy, parallel execution, and async compute.
- Experience profiling and optimizing GPU workloads.
Programming
- Strong proficiency in C/C++.
- Experience developing system -level software and performance -critical components.
AI/ML
- Good understanding of AI/ML fundamentals.
- Experience optimizing workloads related to:
- Computer Vision (CV)
- Large Language Models (LLMs)
Preferred Skills
- Experience with IREE, TVM, XLA, LLVM, or other MLIR -based deployment frameworks.
- Understanding of compiler/runtime architecture and hardware abstraction layers.
- Knowledge of quantization, mixed -precision inference, and model optimization.
- Exposure to multi -target compilation across CPU, GPU, and NPU.
- Familiarity with GPU profiling tools such as ARM Streamline, Intel VTune, Snapdragon Profiler, etc.
- Contributions to MLIR, LLVM, IREE, or related open -source projects.
Qualification
- Bachelor's/Master's/PhD in Computer Science, Electronics, Engineering, Information Systems, or related discipline.
- 4+ years of relevant experience in compiler development, GPU programming, AI acceleration, or systems engineering.