Agentic AI Engineer (AVP)
Citigroup
- Location
- Tampa Florida United States
- Work model
- On-Site
- Level
- Mid
- Posted
- Aug 21, 2026
Skills
About this role
About Citi Citi, the leading global bank, has approximately 200 million customer accounts and does business in more than 160 countries and jurisdictions. Citi provides consumers, corporations, governments, and institutions with a broad range of financial products and services, including consumer banking and credit, corporate and investment banking, securities brokerage, transaction services, and wealth management. As a bank with a brain and a soul, Citi creates economic value that is systemically responsible and in our clients' best interests. As a financial institution that touches every region of the world and every sector that shapes your daily life, our Enterprise Operations & Technology teams are charged with a mission that rivals any large tech company. Our technology solutions are the foundations of everything we do from keeping the bank safe, managing global resources, and providing the technical tools our workers need to be successful to designing our digital architecture and ensuring our platforms provide a first-class customer experience. We reimagine client and partner experiences to deliver excellence through secure, reliable, and efficient services. Our commitment to diversity includes a workforce that represents the clients we serve from all walks of life, backgrounds, and origins. We foster an environment where the best people want to work. We value and demand respect for others, promote individuals based on merit, and ensure opportunities for personal development are widely available to all. Ideal candidates are innovators with well-rounded backgrounds who bring their authentic selves to work and complement our culture of delivering results with pride. If you are a problem solver who seeks passion in your work, come join us. We'll enable growth and progress together.
About the Team
The Agentic AI Engineer is a transformative professional operating at the intersection of cutting-edge artificial intelligence and enterprise-grade financial technology. This is not a role for those content with the status quo — it is a position for builders, innovators, and applied AI practitioners who are energized by the challenge of turning the immense promise of foundation models into reliable, real-world impact. As Citi continues to evolve its Controls Technology capabilities, this role sits at the heart of that ambition: architecting retrieval-grounded, context-aware, and increasingly autonomous AI systems that meet the exacting standards of one of the world's most complex financial institutions. The successful candidate will work shoulder-to-shoulder with senior developers, AI architects, and business stakeholders across Citi's global organization, driving the development of agentic AI applications that are not only technically sophisticated but genuinely transformative. From engineering robust RAG pipelines and knowledge graphs to designing multi-agent workflows and deploying production-grade GenAI applications, this individual will be a driving force behind solutions that enhance operational integrity, accelerate decision-making, and position Citi at the forefront of responsible AI adoption. The impact of your work will extend far beyond lines of code — it will be felt across teams, products, and ultimately, the clients and communities Citi serves around the world.
Responsibilities
Build and integrate generative AI applications using pre-trained and hosted foundation models (via managed GenAI APIs and open-model endpoints). Design and implement context engineering workflows — assembling system instructions, retrieved knowledge, tool definitions, conversation memory, and task metadata into reliable, token-efficient prompts. Contribute to prompt engineering (zero-shot, few-shot, chain-of-thought, role-based prompting) and the development of AI-powered workflows. Develop and maintain Retrieval-Augmented Generation (RAG) systems, including chunking, embedding, hybrid (semantic + keyword) search, and