ML Platform Engineer: Scalable GPU + Kubernetes

Mistral; Palo Alto, Northern (CA, KY); Hybrid; USD 180,000 - 240,000; Full time

Benefits offered by this job

Healthcare coverage

Parental leave

Relocation support

Wellness programs

Meal and transportation allowances

Job summary

Mistral in Palo Alto is seeking a backend-focused ML platform engineer to build and operate the infrastructure powering large-scale training, evaluation, and batch inference.

You will orchestrate GPU workloads, manage compute capacity across clusters, and create self-service workflows to simplify distributed workloads for researchers. This role requires deep Kubernetes knowledge and proficiency in Python or Go, plus GPU tech like PyTorch and CUDA in a fast-moving frontier AI environment.

Qualifications

• 4+ years of experience in ML infrastructure or distributed systems.
• Proficient in Python or Go and production-grade distributed systems.
• Strong Kubernetes knowledge: controllers, scheduling, networking, storage, and resource management.
• Experience with GPU workloads and technologies such as PyTorch, CUDA, NCCL.

Responsibilities

• Build ML Platform APIs, services, and tooling for training, evaluation, fine-tuning, and batch inference.
• Orchestrate GPU workloads with queues, admission control, quotas, priorities, preemption, and topology-aware placement.
• Manage compute capacity across clusters and optimize heterogeneous GPU resource provisioning.
• Enable multi-cluster execution based on capacity, data locality, hardware requirements, and priorities.
• Improve researcher experience with self-service workflows for distributed workloads to launch, observe, debug, and reproduce.
• Improve GPU utilization, scheduling latency, startup times, throughput, and infrastructure efficiency.
• Build for reliability with observability, failure recovery, capacity planning, and operational tooling.
• Operate what you build by participating in on-call rotations and troubleshooting across components.

Skills

Python

Go

Kubernetes

Distributed systems

GPU infrastructure

Tools

Kueue

Karpenter

Volcano

Kyverno

PyTorch

CUDA

NCCL

Job description

Mistral in Palo Alto is seeking a backend-focused ML platform engineer to build and operate the infrastructure powering large-scale training, evaluation, and batch inference.

You will orchestrate GPU workloads, manage compute capacity across clusters, and create self-service workflows to simplify distributed workloads for researchers. This role requires deep Kubernetes knowledge and proficiency in Python or Go, plus GPU tech like PyTorch and CUDA in a fast-moving frontier AI environment.