Member of Technical Staff - GPU Performance Engineer

About the role

Liquid AI, spun out of MIT CSAIL, builds general-purpose AI systems that run efficiently across various deployment targets, including data center accelerators and on-device hardware. We focus on low latency, minimal memory usage, privacy, and reliability, partnering with enterprises in consumer electronics, automotive, life sciences, and financial services. We are currently scaling rapidly and seek exceptional individuals to join our team.

The Opportunity

We are seeking a highly skilled individual to work on performance work related to our novel model architectures. This role involves designing and shipping custom CUDA kernels, profiling at the hardware level, and integrating research ideas into production code that delivers measurable speedups in real-world pipelines (training, post-training, and inference).

The team is small, fast-moving, and highly-ownership-focused. We are looking for someone who finds joy in memory hierarchies, tensor cores, and profiler output. While San Francisco and Boston are preferred, we are open to other locations.

What We're Looking For

• Works profiler-first: Uses tools like Nsight Systems / Nsight Compute to find bottlenecks, validate hypotheses, and iterate until improvements show up in end-to-end benchmarks.
• Bridges theory and practice: Translates ideas from papers into implementations that are robust, testable, and performant.
• Executes independently: Given an ambiguous bottleneck, drives from profiling to kernel/integration changes to benchmarked results to maintained ownership.
• Cares about the details: Focuses on memory hierarchy, occupancy, launch configs, tensor core utilization, bandwidth vs compute limits.

The Work

• Write high-performance GPU kernels for our novel model architectures.
• Integrate kernels into PyTorch pipelines (custom ops, extensions, dispatch, benchmarking).
• Profile and optimize training and inference workflows to eliminate bottlenecks.
• Build correctness tests and numerics checks.
• Build/maintain performance benchmarks and guardrails to prevent regressions.
• Collaborate closely with researchers to turn promising ideas into shipped speedups.

What We're Looking For

• Must-have: Authored custom CUDA kernels (not just calling cuDNN/cuBLAS); strong understanding of GPU architecture and performance; proficiency with low-level profiling (Nsight Systems/Compute) and performance methodology; strong C/C++ skills.
• Desired: CUTLASS experience and tensor core utilization strategies; Triton kernel experience and/or PyTorch custom op integration; experience building benchmark harnesses and perf regression tests.

Success Metrics (Year One)

• Measurable improvement on at least one critical end-to-end pipeline (throughput and/or latency), validated by repeatable benchmarks.
• At least one research-driven technique shipped as a production kernel and maintained over time.
• Performance regressions are detectable early via benchmarks/guardrails, not discovered late.

What We Offer

• Unique challenges: Architectural innovations and efficiency requirements offer unique optimization challenges. High ownership from day one.
• Compensation: Competitive base salary with equity in a unicorn-stage company.
• Health: Pays 100% of medical, dental, and vision premiums for employees and dependents.
• Financial: 401(k) matching up to 4% of base pay.
• Time Off: Unlimited PTO plus company-wide Refill Days throughout the year.