Member of Technical Staff - AI Training Platform
Originally posted September 2, 2026 by the employer.
About the role
As a Member of Technical Staff, AI Training Platform, you will contribute to the infrastructure that powers the model training ecosystem, building and scaling the end-to-end tooling required to train, evaluate, and benchmark models.
What you'll do
- Architect, scale, and maintain the core AI/ML/RL training platform and infrastructure, designing multi-node distributed training systems with elastic sharding, data streaming pipelines, and model checkpointing and recovery.
- Maintain and expand the proprietary training framework to provide a flexible foundation for internal model training.
- Develop and optimize kernels using low-level programming models like CUDA and Triton, designing benchmarking suites to track Model Flops Utilization (MFU), memory bandwidth, and convergence stability.
- Build and iterate on tooling to enable high-level evaluation of novel hardware concepts against benchmarks.
- Translate algorithmic trade-offs with theorists to convert model requirements into concrete specifications for infrastructure and hardware engineering teams.
- Design systems to track and visualize quality-vs-efficiency Pareto frontiers to optimize models for performance and energy consumption.
What you'll need
- BS in Computer Science, Physics, Electrical Engineering, or Applied Math.
- 5+ years of experience in AI/ML engineering, with demonstrated ability to map model architectures (such as transformers, Mixture of Experts, diffusion models) to system performance implications.
- Deep expertise in model partitioning across clusters, communication primitives, and parallelism strategies.
- Strong programming skills in Python or C++.
- Proven track record of implementing, debugging, and maintaining production-grade training frameworks such as Megatron-LM, DeepSpeed, Ray, or PyTorch Lightning.
Nice to have
- MS or PhD or equivalent research/project experience in AI/ML or high-performance computing, with publications.
- Experience in training and post-training large-scale LLMs and generative models.
- Experience in one or more of: Kubernetes, GPU kernels and performance tuning, LLM inference, CUDA, Triton, NCCL, vLLM, SGLang, VeRL, or TRL.
Skills: distributed training, CUDA, Triton, Megatron-LM, GPU kernels
First seen in our index 0 days ago, well below the 58-day median for Infrastructure/Platform Software roles.
Infrastructure/Platform Software · Infrastructure Platform
|
Open roles in category
1040
|
Median days since first seen
58 d
|
Category median salary (all employers)
$236k
|
See the full market breakdown ▾Category comparison, skills in demand, and who else is hiring
| Metric | Unconventional AI | All employers we track in this specialty (1040 roles · 92 employers) |
|---|---|---|
| Open roles in this specialty | 1 | 1040 |
| Open roles in the wider Software, Firmware & Systems family | 6 | 6731 · 166 employers |
| Median days since first seen | 0 d | 58 d (+58 d vs this employer) |
| Median salary (USD postings) | n/a | $236k |
Skills observed across this category: distributed training, CUDA, Triton, Megatron-LM, GPU kernels
Who's hiring in this category
- NVIDIA · 353 open roles · median 63 d
- Qualcomm · 65 open roles · median 59 d
- AMD · 55 open roles · median 38 d
- Graphcore · 47 open roles · median 120 d
- Arm Holdings · 38 open roles · median 17 d
- Cerebras · 36 open roles · median 109 d
How we counted: 1040 open Infrastructure/Platform Software (Infrastructure Platform) roles from 92 employers tracked in the SemiconductorJobs index, counted 14 Nov 2026. Specialty figures count only roles carrying this exact specialty label, so an employer's related work in neighbouring specialties is not included there, and is counted in the wider Software, Firmware & Systems family row. Days are counted from when our index first saw each posting, not from the employer's original posting date.