About the role
Specialist agent developer focused on Model Context Protocol implementation, MCP servers, tool integration, secure context exchange, and streamable transports for agentic workflows.
Responsibilities:
- Design and implement MCP servers, tools, resources, prompts, and context-exchange patterns for agent workflows.
- Build secure tool integrations that connect agents to enterprise APIs, knowledge systems, developer tools, and workflow services.
- Implement streamable transports such as HTTP, SSE, and WebSocket for real-time agent-tool communication.
- Define MCP usage standards, schemas, tool contracts, authentication, authorization, and observability patterns.
- Validate tool-calling accuracy, error handling, retries, permissions, rate limits, and safe execution boundaries.
- Collaborate with backend, platform, and agent teams to operationalize reusable MCP services and integrations.
Requirements:
- Bachelor's or master's degree in computer science, AI, engineering, or a related discipline.
- 6+ years of total experience, including 2+ years in agent development, enterprise integration, or MCP-based tool integration.
- Strong Python programming expertise.
- Hands-on experience building MCP servers, tools, resources, prompts, and agent-tool integration patterns.
- Experience with API integration, authentication, authorization, streaming transports, tool contracts, and secure execution boundaries.
- Familiarity with LangGraph / LangChain, LLM APIs, observability, retries, error handling, and AI guardrails.
Mandatory Tech Stack:
- Language: Python.
- Agent Frameworks: LangGraph / LangChain or equivalent agent-orchestration frameworks.
- LLM APIs: OpenAI / Anthropic / Bedrock / Azure OpenAI or equivalent foundation-model APIs.
- MCP: MCP server implementation, MCP tool integration, resources, prompts, schemas, and secure context exchange.
- Transports: Streamable HTTP, SSE, WebSocket, request/response, and streaming patterns.
- Integration and Security: REST APIs, authentication, authorization, tool permissions, rate limits, retries, and observability.