AI Engineer: Post-Training
Core
Own the product capability for improving LLMs via post-training methods (SFT, RLHF, distillation, synthetic data) for industrial engineering workflows.
Role type
Senior hands-on product ML engineer (post-training)
Builds
Production-ready model improvement pipelines and managed workflows for customer data
Domain
Industrial engineering, aerospace & defence, automotive, semiconductors
Deliverable
production ML models
Required skills
Python, PyTorch, distributed training, post-training methods (RLHF, distillation, SFT), synthetic data pipelines, reward modeling, model evaluation, data governance
Preferred skills
Tool use/agent workflows, large-scale rollout generation, regulated/air-gapped environments
Technologies
PyTorch, Python
Responsibilities
Build reliable post-training harnesses covering data preparation to serving; Improve open models against PhysicsX benchmarks; Develop synthetic-data and expert-data pipelines; Design reward signals and detect reward hacking; Ensure customer data remains within customer platform environments; Set product roadmap for model improvement
