ml engineer in agentic commerce

JPMorganChase; Greater London; Presencial; GBP 90.000 - 150.000; Jornada completa

Descripción de la vacante

J.P. Morgan ищет специалиста для проектирования и выпуска высоконагруженных агентов на платформе NEO. Вы будете владеть агентами от прототипа до продакшна и строить надежные системы поиска и обработки знаний с использованием Graph RAG и векторного поиска.

Кандидат обладает опытом разработки решений на Python, знанием ML и статистики, а также опытом работы в облачных средах и Kubernetes. Присоединяйтесь к глобальной команде для реализации передовых банковских и финтех-решений.

Formación

• MS in CS/Math/ML or эквивалентный опыт.
• Практический опыт разработки production LLM/агентских приложений, включая трассировку, оценки и guardrails.
• Сильные навыки Python и знания структур данных.
• Опыт работы с Kubernetes (AWS EKS) и облачными платформами.

Responsabilidades

• Design and ship production agents on NEO, owning them from prototype through production
• Build robust retrieval systems using Graph RAG, knowledge-graph traversal, vector search, chunking, ranking, and grounding strategies
• Design agent memory, including episodic and semantic memory nodes, recall, summarization, and decay policies
• Deploy and operate solutions on public cloud with SDLC, security, resiliency, and observability practices

Conocimientos

Python

Data structures

Machine learning

Information retrieval

Statistics

Kubernetes

SQL/NoSQL

Communication

Cloud platforms

Educación

MS in CS/Math/ML

Herramientas

Databricks

SageMaker

MLFlow

Graph RAG

Descripción del empleo

Описание:

J.P. Morgan is a global financial services provider offering strategic advice and products to corporations, governments, wealthy individuals, and institutional investors. Its Commercial & Investment Bank provides banking, markets, securities services, and payments, while offering strategic advice, raising capital, managing risk, and extending liquidity in markets worldwide.

Задачи:

Design and ship production agents on NEO, owning them from prototype through production Build robust retrieval systems using Graph RAG, knowledge-graph traversal, vector search, chunking, ranking, and grounding strategies Design agent memory, including episodic and semantic memory nodes, recall, summarization, and decay policies Manage organizational context by assembling entitlement-, lineage-, and tenant-aware context for secure agent reasoning Compose multi-agent workflows using A2A and integrate tools and data through MCP servers Build and run task-level and end-to-end agent evaluations, regression suites, LLM-as-judge, and quality and safety gating Deploy and operate solutions on public cloud with SDLC, security, resiliency, and observability practices Partner with product and business teams to turn use cases into shipped, supported agents Build traditional ML model training pipelines and productionize them using MLOps best practices Develop batch and online inference for ML models

Требования:

MS in Computer Science, Statistics, Mathematics, Machine Learning, or a related field, or equivalent experience Hands-on experience building production LLM-powered or agentic applications, including tracing, evaluations, and guardrails Strong Python programming skills and deep knowledge of data structures, algorithms, machine learning, data mining, information retrieval, and statistics Knowledge of Kubernetes (AWS EKS) Experience training models in Databricks and SageMaker Experience with MLFlow Practical RAG experience with retrieval quality, embeddings, and vector stores Expert knowledge of at least one of AWS, Azure, or Kubernetes Knowledge of data management and data model design, and real-time processing using SQL and NoSQL stores Excellent communication skills and ability to partner effectively with senior technical and business stakeholders Будет плюсом: Graph RAG, agent frameworks or runtimes, A2A, MCP, agent memory design, organizational context management, knowledge graphs and graph databases for retrieval, LLM fine-tuning, small language model inference, full-stack development with modern JavaScript/TypeScript frameworks for agent UIs, experience in the financial or payments domain at a large institution

Условия:

Локация: Лондон