Lead Python Engineer — MCP Protocol & Tool Integration

EPAM Systems; United States; À distance; USD 140 000 - 190 000; Plein temps

Résumé du poste

EPAM Systems seeks a Lead Python Engineer to build MCP protocol servers and agent tool integrations within an Enterprise Agent Development Platform on AWS. You will design and implement scalable, cloud-native components that enable AI agents to be defined, orchestrated, deployed, and observed at scale.

Responsibilities include building MCP servers/clients with async Python, enabling tool integrations, translating APIs to MCP interfaces, and ensuring robust security and governance.

Qualifications

• 5+ years of Python engineering experience with agent/AI systems.
• Hands-on experience implementing MCP servers or clients in production.
• Portfolio of tool integrations for LLM agents (OpenAI functions, LangChain tools, etc.).
• Experience building async HTTP services and streaming protocols.

Responsabilités

• Build MCP protocol servers supporting Streamable HTTP transport, tools/list, tools/call, and capability negotiation.
• Create MCP server/client components using Python async programming and streaming responses.
• Develop agent tool integrations with function calling, tool schema definitions, and parameter validation.
• Set up MCP target types, including Lambda functions as tools and remote MCP servers.
• Handle tool metadata management: schema registration, tagging, and versioning.
• Register targets in AWS AgentCore Gateway for tool discovery.
• Convert existing APIs into MCP-compliant interfaces.
• Collaborate with teams to standardize tool integration patterns.

Connaissances

Python

Async programming

Agent/AI systems

AWS Lambda

MCP protocol

OpenAPI to MCP

LangGraph / LangChain

Outils

OpenAI functions

LangGraph tools

AWS Gateway

Description du poste

We're seeking a Lead Python Engineer to build MCP protocol servers and agent tool integrations as part of an Enterprise Agent Development Platform, a cloud-native, production-grade ecosystem designed to help engineering teams define, orchestrate, deploy, and monitor AI agents at scale. Using LangGraph and Strands Agents on AWS AgentCore Runtime, this platform brings consistency to agent development across the organization. The project covers agent framework design, runtime architecture, marketplace integration, CI/CD automation, and enterprise observability, cutting agent development time from months to days while maintaining strong security, quality, and governance standards.

Responsibilities

• Build MCP protocol servers that support Streamable HTTP transport, tools/list, tools/call, and capability negotiation
• Create MCP server and client components leveraging Python async programming, async HTTP, and streaming responses
• Develop agent tool integrations using function calling, tool schema definitions, and parameter validation patterns
• Set up MCP target types, including Lambda functions as tools, remote MCP servers, and OpenAPI-to-MCP translation
• Handle tool metadata management, covering schema registration, capability tagging, and versioning
• Register targets in AWS AgentCore Gateway to enable agent tool discovery
• Convert existing APIs into MCP-compliant tool interfaces
• Work with engineering teams to establish consistent tool integration patterns across the platform

Requirements

• 5+ years of Python engineering experience working with agent/AI systems
• Practical experience implementing MCP (Model Context Protocol) servers or clients in a production agentic AI setting, such as handling tools/list and tools/call, registering targets with AWS AgentCore Gateway, or performing Lambda-as-tool / OpenAPI-to-MCP translation
• A portfolio of tool integrations developed for LLM agents, such as OpenAI functions, LangChain tools, or similar
• Proficiency in building async HTTP services and streaming protocols
• Strong expertise in AWS Lambda development
• English proficiency at B2 level or above
• Nice to have Prior experience implementing the MCP protocol (server or client side)
• Familiarity with LangChain / LangGraph tool integration
• Experience with AWS Bedrock AgentCore Gateway target registration
• Understanding of semantic tool indexing, including generating embeddings for tool descriptions
• Exposure to agent framework integrations such as Strands, Crew, or AutoGPT