Principal Machine Learning Engineer

United States of America

Full Time

4 hours ago

Visa Sponsor

Key skills

Deep learningTransformer-based architecturesTraining large-scale machine learning modelsFine-tuning machine learning modelsDeploying machine learning models into productionPyTorchJAXDistributed trainingDeepSpeedFSDPMegatronZeRORayGPU workload optimizationMixed precisionQuantisationMemory-efficient techniquesLLM inference frameworksvLLMTensorRT-LLMFasterTransformerReinforcement Learning from Human Feedback (RLHF)PPOORPOCUDAGPU kernelsScientific computingCompiler optimisationMultimodal modelsDiffusion modelsLarge-scale synthetic data generationApache ArrowSparkOpen-source machine learning contributionsAIMachine LearningMLDeep LearningLLM

About this role

USA Tech Recruit is a rapidly growing AI company focused on building intelligent personal productivity software. The Principal Machine Learning Engineer will define the architecture of the company's core machine learning platform and work on solving large-scale technical challenges.

Responsibilities:

Architect and evolve scalable machine learning systems spanning training, inference, evaluation and deployment
Design and optimise distributed GPU training pipelines for large-scale foundation models
Build high-performance inference platforms that balance latency, throughput, reliability and infrastructure cost
Develop robust evaluation and data pipelines to improve model quality, safety and production performance
Collaborate closely with product, backend and infrastructure teams to integrate AI capabilities into customer-facing applications
Drive continuous improvements across production systems through monitoring, optimisation and rapid iteration

Requirements:

Strong industry experience with deep learning and transformer-based architectures
Hands-on experience training, fine-tuning or deploying large-scale machine learning models into production
Expert knowledge of PyTorch, JAX, or similar modern machine learning frameworks
Experience with distributed training and inference technologies such as DeepSpeed, FSDP, Megatron, ZeRO, or Ray
Strong software engineering skills with experience building production-grade, maintainable systems
Experience optimising GPU workloads through mixed precision, quantisation and memory-efficient techniques
Ability to design and deliver complex ML systems from concept through production in fast-paced environments
Experience with LLM inference frameworks including vLLM, TensorRT-LLM or FasterTransformer
Experience with RLHF techniques including PPO, DPO or ORPO
Background in CUDA, GPU kernels, scientific computing or compiler optimisation
Experience with multimodal models, diffusion models or large-scale synthetic data generation
Knowledge of distributed data processing technologies such as Apache Arrow, Spark or Ray
Contributions to open-source machine learning or systems software