Senior Machine Learning Engineer, ML Infrastructure- Online

Unity Technologies•Seattle, WA

1d•$187,200 - $243,300

About The Position

We are seeking a Senior ML engineer to design and evolve Unity Vector’s online model inference platform. This role focuses on building reliable infrastructure for serving machine learning models in production, optimizing inference performance, and enabling safe, efficient experimentation across high-traffic online systems. You will work closely with ML engineers, platform teams, and product stakeholders to ensure models can be deployed, scaled, monitored, and iterated on efficiently. You will play a key role in shaping how models are packaged, served, validated, monitored, and optimized in production environments. This role requires strong systems thinking, deep experience with production ML infrastructure, and the ability to drive architectural improvements across teams.

Requirements

Experience building and operating production-grade online ML inference systems, such as NVIDIA Triton Inference Server, TorchServe, Ray Serve, TensorFlow Serving, or similar systems.
Experience with model serving frameworks such as NVIDIA Triton Inference Server, TorchServe, Ray Serve, TensorFlow Serving, or similar systems.
Experience optimizing inference workloads using techniques such as dynamic batching, model compilation, quantization, GPU acceleration, GPU kernel optimization, caching, or runtime tuning.
Strong experience with distributed systems, Kubernetes, autoscaling, service reliability, and production observability.
Strong programming skills in Python, with practical experience working on production ML systems and high-scale services.
Experience with PyTorch and modern model deployment workflows, including model packaging, validation, and serving lifecycle management.
Experience designing infrastructure for safe model rollout, canary testing, A/B experimentation, and automated rollback.
Strong systems thinking, with the ability to reason about latency, throughput, reliability, scalability, and cost tradeoffs in online systems.
Proven ability to lead technical direction and influence architectural decisions across teams without formal authority.

Responsibilities

Design and operate large-scale online inference infrastructure that serves production ML models with low latency and high reliability, such as PyTorch, Triton Inference Server, Kubernetes, GKE, Ray, or similar distributed serving frameworks.
Develop infrastructure that supports distributed training workflows using technologies such as Pytorch, Ray Data, and Ray Train, etc.
Integrate ML pipelines with workflow orchestration systems (e.g., Flyte, Airflow, or similar) to enable reliable multi-stage training workflows
Optimize model performance through model compilation, GPU/CPU utilization improvements, request scheduling, kernel fusion, and runtime-level tuning.
Improve observability of ML systems through latency, throughput, error-rate, cost, saturation, and model-health monitoring.
Partner closely with ML engineers to support faster model iteration while maintaining production safety, scalability, and cost efficiency.
Improve the reliability and reproducibility of model serving workflows, including model packaging, artifact validation, compatibility testing, and deployment automation.
Lead architectural improvements that make the online ML platform more robust, user-friendly, scalable, and cost-efficient.