Job Description
We are seeking a hands-on AI Engineer to design, build, and deploy production-grade agentic applications leveraging Claude (or similar agentic coding tools) and the Model Context Protocol (MCP). This role goes beyond prompt engineering and prototypes—you will build real agents that act on real systems, integrating enterprise data, tools, and workflows.
You will work across agent orchestration, MCP servers, agent-to-tool interfaces, and cross-platform agent migration (e.g., Claude ↔ Copilot Studio), enabling scalable, secure, and maintainable AI systems embedded directly into business operations. This role requires strong Python engineering skills, deep understanding of LLM tool use and multi-step cognition, and practical experience deploying agentic systems in production environments.
Key Responsibilities
1. Agentic Application Development (Claude)
• Design and implement multi-step agentic workflows using Claude with advanced tool-use, reasoning loops,
and prompt chaining
• Build agents that operate against live enterprise data and systems, not static demos or isolated notebooks
• Implement agent patterns including:
o Planner / Executor architectures
o Tool-selection and dynamic routing
o Reflection, retries, and guardrail handling
• Use the Anthropic Python SDK and Claude Skills / Plugins framework to create maintainable, extensible agent codebases
• Optimize prompt and tool design for reliability, cost, latency, and correctness
2. Claude Cowork Agent & Connector Development
• Build and configure Claude Cowork agents, connectors, and scheduled workflows for enterprise use cases
• Extend Cowork using custom plugins and connectors to integrate domain-specific systems (internal APIs, data stores, BI tools, etc.)
• Define safe execution boundaries, permissions, and automation triggers for unattended or scheduled agent runs
• Partner with product, data, and security teams to ensure agents are safe, auditable, and production-ready
3. Cross-Platform Agent Migration & Portability
• Port agent logic, prompt libraries, and tool integrations across LLM platforms (e.g., Claude ↔ Microsoft Copilot Studio)
• Identify:
o What agent components are portable (reasoning structure, tool semantics)
o What must be rebuilt due to platform constraints (auth, execution models, UI, orchestration patterns)
• Refactor agent designs to satisfy different platform limits around memory, tools, grounding, and execution
• Maintain architectural documentation that enables future platform migration with minimal rework
4. MCP Server Design, Build & Integration
• Design and implement MCP (Model Context Protocol) servers to expose internal tools and data sources to
Claude
• Build end-to-end MCP servers including:
o Tool definitions and schemas
o Authentication & authorization middleware
o Data transformation and response normalization
o Error handling, rate limiting, and observability
• Integrate MCP servers with enterprise APIs, databases, file systems, and SaaS platforms
• Support both local (stdio) and remote (HTTP/streaming) MCP deployment models
• Ensure MCP servers meet enterprise security and compliance standards
5. Production Engineering & Reliability
• Write clean, testable, production-quality Python code
• Implement logging, metrics, and tracing to observe agent behavior and failures
• Optimize agents for:
o Deterministic outputs where required
o Safe degradation and fallback paths
o Responsible AI requirements (privacy, access control, explainability)
• Collaborate with platform, DevOps, and InfoSec teams on deployment pipelines and runtime governance
Desired Experience & Qualification
Core Technical Skills
• Strong Python proficiency (production backend experience, not scripting only)
• Hands-on experience building agentic AI systems (multi-step, tool-using agents)
• Practical experience with Claude and the Anthropic Python SDK
• Deep understanding of:
o Tool-calling LLMs
o Prompt chaining and memory patterns
o Agent planning vs execution models
MCP & Integration Experience
• Hands-on experience implementing MCP servers end-to-end
• Experience exposing internal tools and data to LLMs via structured APIs
• Understanding of authentication patterns (API keys, OAuth, service tokens) for agent access
Platform & Architecture
• Experience migrating or adapting agent solutions across multiple LLM platforms
• Strong understanding of system boundaries, trust models, and platform constraints
• Ability to reason about long-running agents, failure modes, and retry strategies