
SLM Training Engineer
Rockwell Automation3 days ago
Mayfield Heights, OH, USA +3 moreSenior
Responsibilities
- Train, fine-tune, distill, evaluate, and deploy small language models for Rockwell-specific domains and agent tasks.
- Build transformer-based decision models that produce typed choices, scores, and calibrated probabilities.
- Apply decision models to agent routing, request triage, guardrails, and automated evaluation of LLM and agent outputs.
- Build data pipelines involving data curation, labeling, synthetic data generation, and distillation from larger teacher models.
- Design evaluations measuring accuracy, calibration, latency, and cost across SLMs, decision models, classifiers, and LLMs.
- Optimize models for production, edge, on-premises, and industrial deployment using quantization, pruning, and efficient serving.
- Operate the model lifecycle through experiment tracking, model registries, monitoring, drift detection, calibration monitoring, and retraining.
- Partner with agent, platform, and business teams to identify suitable bounded decisions and make build-versus-buy recommendations.
Requirements
- Bachelor’s degree in computer science, machine learning, data science, statistics, or a related field.
- Legal authorization to work in the United States.
- Typically five or more years of relevant professional experience, including three or more years training and deploying machine learning or NLP models in production.
- Experience fine-tuning, distilling, evaluating, and deploying small language models and decision models.
- Knowledge of probability calibration, uncertainty measurement, model evaluation, and reward-based training for calibrated decisions.
- Experience with Python, PyTorch, GPU training infrastructure, and distributed training.
- Experience with major cloud platforms such as Microsoft Azure, AWS, or Google Cloud Platform.
- Preferred experience includes enterprise model routing, guardrails, LLM evaluation, edge or industrial deployment, reinforcement learning for language models, and coordination across geographically distributed teams.
- Familiarity with System One decision models such as TypeSafe’s Jev, RLCD or similar methods, and Scaled Agile Framework methodologies.
- Master’s degree or PhD in computer science, machine learning, statistics, or a similar discipline is preferred.
Benefits
- Health insurance including medical, dental, and vision coverage.
- 401(k), paid time off, and parental and caregiver leave.
- Flexible work schedule coordinated with the manager.
- Hybrid work arrangement requiring work at a Rockwell location at least Mondays, Tuesdays, and Thursdays unless there is a business obligation outside the office.
Tech Stack
Categories
About Rockwell Automation
Rockwell Automation builds industrial control hardware and factory software for manufacturers, and sells services and systems to design, operate, and maintain automated plants. Its Allen-Bradley controllers, drives, and safety products pair with FactoryTalk software and MES to support discrete and process industries. Founded in 1903 and headquartered in Milwaukee, it is a public company on the NYSE (ROK) with global sales, partners, and OEM customers.