Machine Learning Engineer - Training Optimization

Position Overview

Our partner company based in Ireland is seeking a Machine Learning Engineer specialized in Training Optimization. This role offers a significant opportunity to enhance the core methodologies behind large-scale AI model training and deployment.

Key Responsibilities

• Enhance large-scale model training pipelines to boost throughput, stability, convergence rates, and computational efficiency.
• Advance distributed training strategies including data, model, and pipeline parallelism.
• Fine-tune essential training elements such as optimizers, learning rate schedulers, batch size configurations, and numerical precision techniques (bf16, fp16, fp8).
• Detect and alleviate performance bottlenecks through profiling, system diagnostics, and infrastructure enhancements.
• Work collaboratively with research teams to create architecture-aware training methods to further model performance.
• Develop and support robust training infrastructure featuring checkpointing, fault tolerance, and reproducible workflows.
• Assess and integrate advanced training approaches like gradient checkpointing, ZeRO, FSDP, and bespoke optimization methods.
• Define and monitor training performance metrics to promote ongoing improvements in efficiency.
• Transform research insights into scalable, production-ready systems.

Candidate Requirements

• Demonstrated experience in training large-scale neural networks, including extensive language models or similarly complex systems.
• Practical expertise in machine learning training optimization beyond basic model usage.
• In-depth understanding of backpropagation, optimization algorithms, training behaviors, and convergence processes.
• Experience with distributed training frameworks and large-scale computational environments.
• Proficiency with PyTorch and contemporary machine learning development pipelines.
• Ability to work near hardware boundaries including GPU performance, memory constraints, and network considerations.
• Strong programming skills to implement research concepts into reliable, production-quality code.
• Preferred experience with multi-node and multi-GPU systems.
• Familiarity with platforms such as DeepSpeed, FSDP, Megatron, or equivalent custom stacks is advantageous.
• Experience optimizing workloads on NVIDIA or AMD GPU architectures beneficial.
• Contributions to open-source machine learning projects or infrastructure recognized positively.
• Knowledge of neural network architectures beyond Transformer models is a plus.

Benefits

• Competitive remuneration with equity participation opportunities.
• Engagement with cutting-edge AI models and large-scale training ecosystems.
• High degree of ownership influencing technical trajectory and company expansion.
• Collaboration within a compact, highly skilled engineering and research team focused on innovation and quality.
• Rapid feedback cycles promoting experimentation and impactful contributions.
• Challenging environment addressing complex machine learning infrastructure at scale.
• Commitment to technical excellence, continued learning, and advanced AI development.
• Chance to contribute foundational systems shaping the future landscape of AI.

Additional Information

This listing is managed by a trusted partner company who handles all recruitment processes including application review and interviews.

Privacy and Hiring Process: Candidate applications are processed under applicable data protection regulations. AI tools help facilitate unbiased application matching but final hiring decisions rest with human evaluators.