Tech Lead Manager — LLM Training Platform

Scale AI, Inc.; New York (NY); On-site; USD 264,800 - 331,000; Full time

Benefits offered by this job

Health, dental and vision coverage

Retirement benefits

Learning and development stipend

Generous PTO

Commuter stipend

Job summary

Scale AI, Inc. is seeking engineers to build and optimize the training and inference framework that powers the company’s LLM platform.

You will work closely with ML researchers and data scientists to accelerate research and enable next-generation model development, data curation, and evaluation pipelines. Ideal candidates combine system optimization expertise with experience in multi-node LLM training, distributed ML systems, and post-training methods like RLHF/RLVR.

Qualifications

• Experience with multi-node LLM training and inference.
• Experience with developing large-scale distributed ML systems.
• Experience with post-training methods like RLHF/RLVR and related algorithms like PPO/GRPO.
• Strong software engineering skills with CUDA, PyTorch, transformers, flash attention.

Responsibilities

• Build, profile and optimize our training and inference framework.
• Collaborate with ML and research teams to accelerate their research and development, and enable them to develop the next generation of models and data curation.
• Research and integrate state-of-the-art technologies to optimize our ML system.

Skills

System optimization

Multi-node LLM training

Distributed ML systems

RLHF/RLVR

PPO/GRPO

CUDA

PyTorch

Transformers

Flash attention

Communication skills

Tools

CUDA

PyTorch

Transformers

Flash attention

Job description

Scale AI, Inc. is seeking engineers to build and optimize the training and inference framework that powers the company’s LLM platform.

You will work closely with ML researchers and data scientists to accelerate research and enable next-generation model development, data curation, and evaluation pipelines. Ideal candidates combine system optimization expertise with experience in multi-node LLM training, distributed ML systems, and post-training methods like RLHF/RLVR.