Performance Engineers with deep expertise in low-level systems optimization specifically C++, Python, and Rust to bring hands‑on technical excellence and elevate the quality of our AI training and inference infrastructure data.
Performance Engineer (C++, Python, Rust) [$90-$120/hr] (W2, Contingent Role)
This is a W‑2 employment position.
Role Responsibilities
- Guide research and engineering teams to close knowledge gaps and improve AI model performance in systems‑level optimization, compiler engineering, and runtime performance topics.
- Design challenging, domain‑relevant tasks across multiple specializations, and write accurate and well‑structured solutions to performance engineering and systems programming problems.
- Evaluate performance engineering tasks and solutions and provide clear, written technical feedback.
- Develop guidelines and detailed rubrics/evaluation frameworks to assess systems design quality, profiling methodology, memory optimization, and performance analysis across AI workloads.
- Collaborate with other subject matter experts to ensure consistency and accuracy in training data.
Good Candidature
- 5+ years of dedicated professional experience in performance engineering, systems programming, or low‑level optimization at a recognized, top‑tier organization.
- Deep hands‑on expertise in C++ for performance‑critical systems including memory management, concurrency primitives, SIMD intrinsics, and compiler‑level optimization (e.g., PGO, LTO, custom allocators).
- Proficiency in Python for profiling, benchmarking, and tooling workflows; experience with Rust in systems‑level or safety‑critical performance contexts is a strong plus.
- Demonstrable track record of measurable performance improvements on production systems (e.g., latency reduction, throughput gains, memory footprint optimization).
- Demonstrable career progression.
- Ability to engage reliably for at least 30 hours/week during weekdays.
- Strong written communication skills and the ability to explain complex technical decisions clearly.