Compiler Engineer

Acceler8 Talent is partnered with a venture-backed, stealthy startup to hire a Compiler Engineer.

They're building high-performance AI inference platforms at the rack level, taking an approach that leverages a differentiated system-on-chip architecture to drive efficiency improvements for large-scale inference workloads in data center environments. The team is developing both custom hardware and extending open-source software to support advanced machine learning models.

This role involves contributing to, or leading*, the development of machine learning compiler capabilities within a modern compiler stack. Depending on experience, you will either take ownership of key initiatives or work as part of a small, highly collaborative team. The scope spans the full stack from high-level frameworks down to low-level execution layers targeting GPUs and emerging accelerator hardware.

Key Responsibilities

• Develop and enhance components of an ML compiler stack, focusing on performance and scalability
• Translate workloads from high-level frameworks (e.g., PyTorch, Triton) into efficient low-level representations
• Optimize generated code for execution on GPUs (e.g., CUDA, ROCm) and other AI-focused hardware platforms
• Partner closely with hardware and systems teams to support co-design efforts in a fast-paced environment

Qualifications

• Advanced degree (PhD preferred) with approximately 5 years of relevant experience in compilers, machine learning systems, or related areas, or equivalent practical experience
• Strong foundation in compiler design, along with experience in performance tuning and optimization
• Hands-on experience with ML frameworks and compiler stacks, particularly involving PyTorch, Triton, and low-level code generation
• Familiarity with GPU architectures and experience working across hardware–software boundaries is a plus

*Leveling (Senior vs. Principal) is based on scope of impact, ownership, and leadership responsibilities rather than strictly years of experience