Lead Machine Learning Engineer-MLOps

JPMorgan Chase & Co.; Palo Alto (CA); On-site; USD 180,000 - 240,000; Full time

Job summary

JPMorgan Chase & Co. is seeking a Senior MLOps engineer to partner with Data Scientists to build and deploy ML models on a modern MLOps stack in a Palo Alto setting.

You will lead pipelines for distributed training on GPU clusters, model serving at scale, and continuous validation in a controlled environment. The role focuses on production ML workflows within Personalization and Insights, deploying on AWS, and enabling high-throughput, low-latency applications that power personalized experiences

Qualifications

• BS in Computer Science or related field with 6+ years experience, or MS with 4+ years.
• Strong Python and AWS cloud experience.
• Knowledge of quantization techniques for LLMs.
• Experience with caching, CUDA, autoscaling, high throughput, low latency.
• Data science fundamentals and model training/deployment.
• Experience with monitoring/observability for ML models.
• Familiarity with Ray, DuckDB, Spark and training/inference systems.

Responsibilities

• Build, deploy, and maintain pipelines for distributed training on GPU-enabled clusters.
• Develop and manage real-time and batch inference pipelines.
• Execute quantization and deploy LLMs for efficient inference.
• Optimize vector DBs to support AI apps.
• Establish comprehensive monitoring and observability pipelines.
• Collaborate with cross-functional teams to adopt new tech.
• Partner with product and architecture teams to design scalable solutions.

Skills

Python

AWS Cloud

Quantization PTQ/AWQ

CUDA

Distributed training

Monitoring/Observability

Ray/DuckDB/Spark

vLLM

Education

BS in CS or related field

MS in CS or related field

Tools

Ray

DuckDB

Spark

vllm

Job description

We are looking for a Senior MLOps engineer to work closely with Data Scientists to build and deploy ML models on a modern MLOps stack.

As Lead Machine Learning Engineer on the Recommendation Engine team, you’ll build and maintain pipelines for distributed model training on large compute clusters, batch/real-time model serving, hyperparameter tuning at scale, model monitoring, production validation and other activities vital for model development, testing and deployment in a well-managed, controlled environment.

Our product, Personalization and Insights, builds and supports high throughput, low latency applications which leverage state of the art machine learning architectures, and which are deployed in AWS. These applications power personalized experiences across Chase Consumer & Community Banking channels, to help weave a user experience that includes traditional banking services with other services in the Travel, Merchant Offer Shopping, and Dining spaces.

Job responsibilities

• Build, deploy, and maintain robust pipelines for distributed training on GPU-enabled clusters to support scalable machine learning workflows.
• Develop and manage pipelines for high-throughput, real-time inference as well as batch inference, ensuring optimal performance and reliability.
• Implement quantization techniques and deploy large language models (LLMs) to maximize efficiency and resource utilization.
• Oversee the management and optimization of vector databases to support advanced AI and machine learning applications.
• Establish and maintain comprehensive monitoring and observability pipelines to ensure system health, performance, and rapid issue resolution.
• Collaborate with cross-functional teams to integrate new technologies and continuously improve existing infrastructure.
• Partner with product, architecture, and other engineering teams to define scalable and performant technical solutions.

Required qualifications, capabilities, and skills

• BS in Computer Science or related Engineering field with 6+ years of experience Or MS degree in Computer Science or related Engineering field with 4+ years experience.
• Solid knowledge and extensive experience in Python and in cloud computing, preferably AWS
• Understanding of quantization techniques such as PTQ, AWQ etc. used to quantize LLMs for accelerating inference on specific GPU architectures
• Experience in systems engineering fundamentals: caching, CUDA, autoscaling, high throughput, low latency, x-region resilient applications
• Deep knowledge and passion for data science fundamentals, training and deploying models
• Experience in monitoring and observability tools to monitor model input/output and features stats
• Operational experience in big data/ML tools such as Ray, DuckDB, Spark and in training/inference systems such as Ray, vllm/SGLang
• Solid grounding in engineering fundamentals and analytical mindset

Preferred qualifications, capabilities, and skills

• Experience with recommendation and personalization systems is a plus.
• CUDA experience is a big plus
• Solid fundamentals and experience in containers (docker ecosystem), container orchestration systems [Kubernetes, ECS], DAG orchestration [Airflow, Kubeflow etc]
• Good knowledge of Databases