1 day ago
Lowell, MA, USA +2 moreStaff+
Base Salary
$134k - $191k/yr
Responsibilities
- Design, build, and maintain scalable ML pipelines covering data ingestion, feature engineering, training, evaluation, deployment, and the full model lifecycle.
- Lead pretraining of large transformer models on proprietary datasets and optimize distributed training across GPU clusters.
- Develop and operate fine-tuning pipelines with experiment tracking, hyperparameter optimization, and model versioning.
- Build model-serving infrastructure for low-latency, highly available, cost-efficient production deployments.
- Implement automated ML workflow testing, validation, canary deployments, monitoring, drift detection, performance alerts, and retraining triggers.
- Optimize training and inference through mixed-precision training, quantization, distillation, pruning, and efficient batching.
- Own end-to-end MLOps infrastructure decisions and establish platform-wide architecture and engineering standards.
- Communicate infrastructure decisions, capacity planning, and reliability considerations to internal and external stakeholders.
- Conduct architecture reviews, establish MLOps best practices, and mentor team members on scalable ML systems.
Requirements
- PhD plus 3 years of experience, or a master’s degree plus 5 years of experience, in machine learning engineering, MLOps, or a related discipline.
- Deep expertise in ML infrastructure, distributed training frameworks, model-serving platforms, and orchestration tools.
- Hands-on experience pretraining or fine-tuning transformer models with 1B+ parameters and techniques including LoRA, QLoRA, RLHF, and DPO.
- Experience with cloud ML platforms, GPU infrastructure, and container orchestration.
- Experience with ML experiment tracking and model registry tools.
- Significant experience with big data tools, databases, and scalable architectures.
- Advanced Python programming skills and expertise with ML frameworks and infrastructure-as-code tools.
- Strong understanding of model optimization and efficient inference strategies.
- Proven experience building and operating reliable production ML systems that serve models at scale.
- Commitment to continuous learning and staying current with ML infrastructure and foundation model developments.
Benefits
- Hybrid work arrangement requiring three days in the office.
- Eligible employees may participate in performance-based bonus and restricted stock unit programs.
Tech Stack
Categories
About UKG
UKG builds cloud software for human capital and workforce management—HR, payroll, timekeeping, and scheduling—sold as SaaS to mid‑size and large employers. Formed in 2020 by the merger of Ultimate Software and Kronos, it is privately held by Hellman & Friedman and has dual headquarters in Lowell, Massachusetts, and Weston, Florida. Its platform helps organizations handle pay, compliance, and workforce planning across industries.
