Senior AI Software Engineer - MCP & Agentic Systems

Wells Fargo & Company•Iselin, NJ

23h•$100,000 - $196,000•Hybrid

About The Position

Wells Fargo is seeking a talented and passionate AI Developer to help shape the next generation of Agentic AI systems. In this role, you'll be at the forefront of building Model Context Protocol (MCP) infrastructure — designing and implementing MCP servers and clients that power intelligent, tool-using agents. You'll work closely with AI developers, and product teams to deliver robust, secure, and scalable agent frameworks that drive real-world impact.

Requirements

4+ years of Software Engineering experience, or equivalent demonstrated through one or a combination of the following: work experience, training, military experience, education
3+ years strong proficiency in Python with experience in building production-grade services
2+ years of LLM experience
1+ years of hands-on knowledge of the Model Context Protocol (MCP) — including protocol internals, transport layers, message schemas, and lifecycle management
1+ years of authentication, authorization, and access control patterns (OAuth 2.0, API keys, RBAC, scopes) applied to AI tool exposure
1+ years hands-on experience building long-running agents with LangChain and/or Google ADK
1+ years of demonstrated experience designing skill-based, modular agents with clearly defined capability boundaries
1+ years with RESTful APIs, async programming, and event-driven architectures

Nice To Haves

Experience with agent observability and tracing tooling (e.g. LangSmith, OpenTelemetry, Arize)
Familiarity with tool schema design and structured output validation (e.g. Pydantic, JSON Schema)
Exposure to containerization and deployment of agent services (Docker, Kubernetes, OpenShift)
Knowledge of prompt engineering and LLM evaluation best practices
Understanding of AI safety principles and responsible agent design

Responsibilities

Design, build, and maintain MCP servers using FastMCP, enabling seamless context sharing between models and tools
Implement MCP clients that integrate with LLM-based applications and agent run times
Engineer authorization and access control layers within MCP infrastructure, ensuring secure and policy-compliant tool exposure
Develop long-running autonomous agents capable of executing complex, multi-step workflows using frameworks such as LangChain or Google ADK
Build skill-based agents that decompose capabilities into reusable, composable skills
Own end-to-end delivery of MCP server implementations — from schema design and tool registration through to authorization enforcement and deployment
Build and maintain a library of reusable MCP-compatible tools and skills accessible across agent runtimes
Implement secure access control policies governing which agents and clients can invoke which tools and under what conditions
Architect and develop long-running agent workflows that handle retries, state persistence, error recovery, and graceful termination
Design skill-based agent systems that allow capabilities to be added, versioned, and composed without tight coupling
Build multi-tool orchestration pipelines where agents autonomously select and sequence tools to complete complex user objectives
Write thorough technical documentation, integration guides, and runbooks for MCP infrastructure
Establish and uphold engineering best practices — code reviews, testing, CI/CD, and observability — across the agentic systems stack
Stay current with the rapidly evolving MCP ecosystem and broader agentic AI landscape, bringing new patterns and improvements to the team