Remote job
Research Engineer, Language Models
Job details
About this role
Role overview This full-time research engineering role focuses on improving language models for practical use on personal devices. You will connect model design, training data, and learning objectives with capabilities such as reasoning, tool use, and conversation, then work with systems engineers to assess how experiments perform under deployment constraints. The role is listed as based in San Francisco or remote.
Responsibilities - Design pre-training and post-training experiments to improve model capability, context handling, and computational efficiency. - Build data preparation and training workflows, and study how data quality and sampling affect model behavior. - Create evaluations for instruction following, reasoning, tool use, and multi-turn interaction, investigating failures that aggregate scores can hide. - Use controlled experiments and ablations to test whether improvements are reproducible and distinguish them from evaluation changes or noise. - Work with inference engineers to understand how compression and quantization affect model capabilities. - Document findings and turn successful work into checkpoints, evaluation tools, and reusable methods.
Requirements - Strong Python skills and practical experience with PyTorch or a comparable deep-learning framework. - Experience training, adapting, or deeply investigating language models, with projects that can be discussed in detail. - Sound understanding of optimization, model architectures, and experimental design. - Ability to debug research code, investigate unexpected results, and explain decisions using evidence. - Careful engineering practices and willingness to work across research and deployment.
Nice to have Experience with distributed training, data quality, distillation, model compression, long-context modeling, preference learning, or adaptation to particular tasks and users.