IBM Linux Software Developer AI Accelerator 

India, Karnataka, Bengaluru 

40729117

What you will do:

• Your work will focus on the exploitation of the unique AI hardware acceleration features of the IBM Z platform. This will include additions to low-level code in the Linux operating system, that interacts directly with the accelerator hardware.
• You will also work on enabling standard AI open source frameworks like PyTorch. With your focus being development work, you will also be responsible for validating functionality, handling problem determination, bug fixing, and solving client problems.
• Your attendance at open source community meetings is recommended and supported and client facing activities (e.g. client support, providing lectures at conferences) are possible. You will closely interact with the respective open source communities (“”upstream integration”” of your code) and our distribution partners (Red Hat, SUSE, Canonical).

Required Technical and Professional Expertise

• Proven experience in operating systems development and low-level programming (e.g. C,C++, Assembler)
• Practice in debugging and problem determination.
• More than 3 years’ experience with Linux software development
• Good understanding of machine learning and AI foundations
• Solid knowledge in problem determination methods and debugging tools (e.g. gdb, Valgrind)
• Experience in Source code repository systems (e.g. git), scripting languages (e.g. Python), working with containers (docker/podman) and test automation (e.g. Jenkins) skills
• Highly motivated Individual to work in a dynamic international development team
• Fluent English language skills both verbal and written.

Preferred Technical and Professional Expertise

• Prior contributions to open-source projects or current open-source maintainer status
• Linux kernel development experience
• Knowledge in AI model optimizations and transformer models
• Skills in IBM Z architecture (“”Principles of Operations””), Z assembler and I/O concepts.
• Understand standing of Endianness