We are sharing a specialised full-time consulting opportunity for US-based MLOps and ML systems engineers with production experience in JAX, PyTorch, distributed training infrastructure, and custom GPU kernel development using Pallas or Triton.
This role supports a high-impact generative AI initiative focused on developing and evaluating advanced ML infrastructure tasks for frontier model training. Selected engineers will design technically challenging problems, produce rigorous solutions, assess model-generated outputs, and help establish evaluation standards across training pipelines, distributed systems, framework-level optimisation, and GPU kernel performance.
Key Responsibilities
ML Infrastructure & Training Systems
- Analyse and improve machine learning training infrastructure, deployment workflows, and model-development systems
- Guide research and engineering teams on MLOps, distributed training, and ML framework-level challenges
- Evaluate training-pipeline architecture, scalability, reliability, and performance
- Identify technical gaps affecting model training, experimentation, and infrastructure efficiency
Technical Task & Solution Development
- Design challenging tasks covering MLOps, ML systems, training infrastructure, and framework-level engineering
- Write accurate, technically rigorous, and well-structured solutions
- Develop realistic scenarios involving distributed systems, accelerator utilisation, and production ML workflows
- Ensure tasks reflect practical engineering challenges encountered in advanced AI environments
JAX, PyTorch & GPU Kernel Evaluation
- Evaluate technical work involving JAX and PyTorch at production scale
- Review custom GPU kernels written or optimised using Pallas or Triton
- Assess kernel correctness, memory access patterns, computational efficiency, and hardware utilisation
- Analyse framework-level implementation choices and identify opportunities for performance improvement
Evaluation Frameworks & Technical Feedback
- Compare alternative technical solutions and determine which approach is more accurate and effective
- Provide clear written feedback on correctness, system design, scalability, and optimisation quality
- Develop detailed rubrics for evaluating training pipelines, distributed systems reasoning, and kernel-level implementations
- Collaborate with other technical specialists to maintain consistency across evaluation standards and training data
Ideal Profile
Strong candidates may have:
- At least 2 years of dedicated professional experience in MLOps, ML infrastructure, or ML systems engineering
- Production experience with JAX, PyTorch, or both at meaningful scale
- Hands-on experience writing or optimising custom GPU kernels using Pallas or Triton
- Strong knowledge of model-training pipelines, distributed systems, accelerators, and performance optimisation
- Experience working within a recognised technology, AI research, or high-performance engineering organisation
- Demonstrable professional growth and increasing technical responsibility
- Strong written communication and the ability to explain complex engineering decisions clearly
- Reliable availability for a full-time, 40-hour weekday schedule
Educational Background
- A degree in computer science, machine learning, electrical engineering, applied mathematics, or a related technical field is highly relevant
- Graduate-level education in machine learning systems, distributed computing, or high-performance computing may be helpful
- Equivalent professional experience in production ML infrastructure may also be considered
- Advanced technical work involving GPU programming, compiler systems, or large-scale model training is especially valuable
Nice to Have
- Experience supporting large language model or generative AI training environments
- Familiarity with distributed training frameworks, accelerator orchestration, and multi-host systems
- Knowledge of XLA, CUDA, compiler optimisation, or low-level performance engineering
- Experience benchmarking GPU workloads and diagnosing training-performance bottlenecks
- Familiarity with model-evaluation pipelines, technical annotation, or structured training-data development
- Previous involvement in technical review, engineering mentorship, or rubric development
- Experience collaborating with research scientists and infrastructure engineering teams
Why This Opportunity
- Contribute to advanced generative AI and large-scale model-training initiatives
- Apply deep expertise in JAX, PyTorch, Pallas, Triton, and ML infrastructure
- Work on challenging problems spanning training systems, distributed computing, and GPU optimisation
- Influence the quality of technical training data used in frontier AI development
- Join a full-time remote engagement with competitive hourly compensation
Contract Details
- Full-time W-2 contingent employment arrangement
- Fully remote role available to candidates based in the United States
- Expected commitment of 40 hours per week during weekdays
- This engagement requires full professional availability without conflicting employment or external commitments
- Competitive rates between $65–$105 per hour depending on expertise and project scope
- Immediate availability is preferred
- Work may include onboarding, technical calibration, and ongoing quality-review activities
- Project scope and duration may be adjusted according to programme requirements and performance
About the Platform
This opportunity is available through 24-MAG LLC. We connect experienced professionals with remote consulting opportunities across technical, evaluation, and project-based workstreams.
By submitting this application, you acknowledge that your information may be processed by 24-MAG LLC for recruitment and opportunity matching in accordance with our Privacy Policy: https://www.24-mag.com/privacy-policy.