Senior Agentic AI Engineer

Citizens Bank•Johnston, RI

4h•Hybrid

About The Position

We are seeking a hands-on, results-driven Senior AI Software Engineer to build and scale Agentic AI capabilities—AI systems that can plan, act, and complete tasks through tool use and workflow orchestration—across enterprise platforms. This role is pivotal in delivering secure, reliable, and compliant AI Agents that integrate into customer and colleague journeys, accelerate time-to-market, and improve operational efficiency. You will operate at the intersection of software engineering and applied AI, owning end-to-end delivery—from architecture and implementation to deployment, observability, and operational excellence.

Requirements

Bachelor’s or Master’s degree in Computer Science, Engineering, or related field (or equivalent practical experience).
7+ years building production software systems; 3+ years delivering AI/ML or GenAI solutions to production.
Strong proficiency in Python plus at least one backend language (Java/Go/Node.js).
Demonstrated experience building distributed systems, APIs, microservices, and event-driven architectures in cloud environments.
Hands-on experience with LLM application development (tool calling/function calling, orchestration frameworks, evaluation).
Strong engineering discipline: testing strategy, performance tuning, CI/CD, reliability, and on-call/incident response.

Nice To Haves

Proven experience building AI Agents / Agentic workflows (multi-step task execution, tool integration, planning/execution separation).
Experience implementing RAG for enterprise knowledge grounding and traceability.
Experience with containerization and orchestration (Docker/Kubernetes) and enterprise observability stacks.
Experience in regulated environments (financial services), including SDLC controls, privacy constraints, and audit readiness.

Responsibilities

Design and implement AI Agent architectures (planner/executor patterns, multi-step reasoning flows, tool-use orchestration) to execute business tasks end-to-end.
Build tooling and integrations for agents (APIs, internal services, data tools, workflow engines), including secure action execution, approvals, and guardrails.
Engineer agent behaviors for reliability: deterministic fallbacks, bounded autonomy, retries, idempotency, and safe failure modes for partial completion.
Develop reusable agent frameworks and components: shared tool registry, prompt/tool contracts, agent memory strategies, policy enforcement, and standardized observability.
Enable teams to embed agents into multiple products/channels via SDKs, reference implementations, and well-defined APIs, reducing duplication and improving consistency.
Implement evaluation pipelines for agents: task success rate, tool-call accuracy, groundedness, latency/cost, and regression testing across agent versions.
Build production-grade RAG pipelines and grounding mechanisms to ensure agent outputs are traceable and based on authoritative enterprise knowledge sources.
Implement agent memory patterns (session memory, long-term memory, semantic caches) with strong privacy and retention controls.
Apply prompt orchestration techniques (system policies, role prompts, tool prompts) to reliably guide agent behavior.
Build and maintain production inference services optimized for performance, resiliency, and cost (caching, batching, async execution, circuit breakers).
Create CI/CD for agent services: automated tests, canary releases, rollback strategies, and safe progressive delivery.
Implement full observability: distributed tracing for agent steps and tool calls, model/token telemetry, failure analytics, and incident runbooks.
Embed security-by-design into agent systems: secrets handling, least-privilege tool access, encryption, audit logs, and tamper-resistant action trails.
Implement policy and governance guardrails for agent autonomy: approval flows, restricted actions, PII redaction, and prompt injection defenses.
Partner with Risk/Compliance to ensure solutions meet enterprise governance expectations.
Lead design reviews and mentor engineers on agentic architecture, production readiness, and engineering best practices.
Collaborate with Product, Architecture, Security, and Data partners to translate business tasks into scalable agent workflows and services.