JPMorgan Chase
Lead Software Engineer - Machine Learning
$157K–$215K
Palo Alto, CA +1 · Full-time · Posted Sep 9, 2026
- Location
- Palo Alto, CA
Seattle, WA - Experience
- 5+ years applied experience; no alternatives stated.
- Education
- Formal training or certification on software engineering concepts.
About the job
Builds and operates scalable cloud-based machine-learning training platforms, optimizing GPU workloads and enabling governed GenAI and LLM training workflows.
What you'll do
- Design, build, and maintain end-to-end ML training platforms and Kubernetes-based infrastructure.
- Optimize single-node and distributed GPU training for performance, cost, reliability, and reproducibility.
- Enable GenAI/LLM training and fine-tuning, and improve observability, automation, governance, and developer workflows.
What they ask for
- Formal training or certification in software engineering concepts and 5+ years of applied experience.
- Experience running cloud ML training and debugging infrastructure and code; strong Python engineering practices.
- Hands-on deep-learning training experience with PyTorch or TensorFlow, including distributed training and training performance optimization.
- Experience with ML automation/CI, Kubernetes and AWS; experience leading safe, validated use of approved AI-assisted software development tools.
Nice to have
- Experience with training across multiple cloud platforms and cloud-native high-throughput networking or storage.
- Familiarity with Spark, Ray, Dask, workflow orchestration, model registries, or large-dataset input pipeline optimization.
Pay
Palo Alto,CA $156,750.00-$215,000.00 USD base salary; Seattle,WA $156,750.00-$215,000.00 USD base salary
Tools
Python, PyTorch, TensorFlow, Kubernetes, AWS, EKS, ECR, S3
AI focus
Enables GenAI/LLM training and fine-tuning workflows, including evaluation and governed GPU execution. Leads adoption of enterprise-authorized AI-assisted engineering tools, with validation and responsible-use practices.
Model-written summary · the original posting is authoritative