Conversation Designer, AI and Agents

The Citco Group Limited•Charlotte, NC

About The Position

The market leader. The premier provider. The best in the business. At Citco, we’ve been the front-runner in our field since our incorporation in 1948 led to the evolution of the asset servicing sector itself. This pioneering spirit continues to guide us today as we innovate and expand, push beyond the boundaries of our industry, and shape its future. From working exclusively with hedge funds to serving all alternatives, corporations and private clients, our organization has grown immensely across asset classes and geographies. For us, this progress is a pattern that we’ll only maintain as we move forward, always prioritizing our performance. So for those who want to play at the top of their game and be at the vanguard of their space, we say: Welcome to Citco.

Requirements

4+ years in conversation design, content design, technical writing, knowledge management, or a closely related discipline, ideally including hands-on work with LLM- or chatbot-based systems.
Demonstrable ability to take dense, expert-written material and restructure it without losing accuracy.
Practical understanding of how retrieval-augmented generation works and why content structure, chunking and metadata affect answer quality. You don't need to build the pipeline, but you need to know what makes it fail.
Strong information architecture instincts: taxonomies, controlled vocabularies, content models.
Excellent written English and a high tolerance for editorial precision.
Proven stakeholder management across functions, including the ability to push back on senior SMEs and chase people who have not prioritised your review request.
Comfort working in a regulated environment with documented approvals and version control.

Nice To Haves

Experience in financial services, ideally fund administration, asset servicing, or another operationally complex, multi-jurisdictional business.
Familiarity with prompt engineering and with evaluating LLM outputs systematically rather than anecdotally.
Working knowledge of Markdown, YAML/JSON metadata, Git-based content workflows, or a headless CMS.
Experience running a content governance model with distributed contributors.
Basic data literacy: able to pull and read usage and evaluation metrics without waiting on an analyst.

Responsibilities

Run the intake process for knowledge contributions from teams across Citco, from initial request through to publication.
Maintain and evolve Citco’s knowledge base content standard, including the style guide, templates, and controlled vocabulary.
Rewrite, restructure and normalize SME-supplied material into the house content standard: consistent article structure, headings, terminology, length, and voice.
Resolve conflicts and duplication where two teams have documented the same process differently, escalating to the relevant owners where a genuine business decision is needed.
Design and maintain the taxonomy, tagging and metadata schema for the KB collection: business line, category, sensitivity classification, review date, content owner etc.
Structure content for retrieval: appropriate chunking, self-contained sections, disambiguating titles, explicit scoping ("this applies to Luxembourg-domiciled funds only").
Own the publishing workflow: staging, review, approval, versioning, and mounting content into the live KB collection consumed by downstream platforms.
Manage deprecation and retirement so stale guidance does not silently keep answering questions.
Define how assistants behave: scope boundaries, tone, response length and format, how to handle partial knowledge, when to hedge, and when to hand off to a human.
Contribute to maintenance of system prompts, response templates, refusal and escalation language across the platforms the KB serves.
Build and maintain evaluation sets: representative questions with agreed-correct answers, used to test the KB before and after changes.
Review conversation logs and user feedback to find where the assistant is wrong, unhelpful, or answering outside its remit, and trace failures back to their root cause: missing content, badly structured content, retrieval failure, or prompt design.
Report on coverage, accuracy, deflection and content freshness to stakeholders.
Work with Risk, Compliance, Legal and Information Security to ensure published content meets regulatory, confidentiality and client-data obligations, and that the approval trail is auditable.
Ensure no client-identifying material enters a general-access knowledge collection.
Act as the day-to-day relationship owner with contributing teams: chasing reviews, training SMEs on how to write for AI retrieval, and making content submission as low-friction as possible.