AI Accelerator Kernel Engineer

Turiyam AI; Bengaluru; On-site; INR 4,000,000 - 7,000,000; Full time

Benefits offered by this job

Competitive salary

Cutting-edge AI tech

Collaborative environment

Growth opportunities

Job summary

Turiyam AI is building world-class GenAI semiconductor solutions from India for global markets. We seek engineers to develop and tune deep learning operations on a next-generation AI hardware, in a highly parallel, heterogeneous environment.

You will design and optimize math libraries for AI applications and work across hardware/software teams to maximize performance. The ideal candidate has 5+ years in high-performance math libraries, GPU/accelerator programming, and a solid grasp of HW

Qualifications

• Bachelor's/Master's/PhD in CS, Math, Computer Engineering or related field.
• 5+ years of proven experience developing high-performance math libraries.
• Experience with GPU or AI accelerator programming.

Responsibilities

• Design, develop, optimize and validate math libraries for AI applications; focus on dense linear algebra and matrix operations.
• Optimize performance of math libraries for hyper-optimized solutions.
• Collaborate with hardware and software teams to integrate libraries with the SW stack.
• Stay updated with AI and math library advancements; research new techniques for accelerators.

Skills

Math libraries

GPU programming

Parallel programming

HW architecture

Education

Bachelor's/Master's/PhD in CS/Math/CE

Tools

CUDA

TensorRT

Job description

At TuriyamAI, we are pioneering world leading GenAI semiconductor solutions from India, for India and the World. Our breakthrough solutions are set to redefine the future of AI computing, driving unparalleled efficiency, performance, and accessibility for enterprises worldwide.

Job Description:

We are looking for an exceptional engineers to develop and tune deep learning operations on highly parallel and custom instruction set of our next‑generation AI hardware. The ideal candidate will have a strong background in heterogeneous computing environment, and experience with parallel programming, distributed algorithms to maximize hardware utilization at rack scale.

Responsibilities:

• Design, develop, optimize and validate math libraries for AI applications, focusing on dense linear algebra, matrix operations, and other relevant mathematical functions. These libraries should be highly efficient and scalable across variety of multi-modal AI models.
• Work on optimizing the performance of AI math libraries to achieve hyper-optimized solutions.
• Collaborate with hardware and software developers, to ensure seamless integration of math libraries with our SW stack
• Stay updated with the latest advancements in AI and math library development. Conduct research to identify new techniques and technologies that can enhance our AI accelerators' performance and efficiency.

Requirements:

• Bachelor's, Master's or Ph.D. degree in Computer Science, Mathematics, Computer Engineering or a related field.
• 5+ years of proven experience in developing high-performance math libraries, preferably for AI applications.
• Experience with GPU or AI accelerator programming
• Experience writing efficient accelerator kernels for matmuls, convolutions, normalisation operators, kernel fusion, etc.
• Understanding of HW architecture and must be comfortable learning new hardware architecture
• Excellent problem-solving skills and ability to work in a blazing-fast-paced environment.

Preferred Qualifications:

• Familiarity with machine learning frameworks such as PyTorch
• Familiarity with compiler technology fundamentals such as: abstract syntax trees, control flows, data flow analysis, IR, and target lowering
• Knowledge of assembly programming, low-level optimizations

What We Offer:

• Competitive salary and benefits package.
• Opportunity to work on cutting-edge AI technology.
• Collaborative and dynamic work environment.
• Professional growth and development opportunities.