AI Agent Manager, Operations

MARCURA EQUITIES FZE

Responsibilities

• Use Case Discovery and Prioritisation: Identify and evaluate high value opportunities to deploy AI agents across operations, including document processing, data extraction, workflow routing, exception handling, scheduling, and compliance checks. Prioritise based on effort, impact, and adoption readiness.

• Agent Design, Build, and Deployment: Design, configure, and deploy AI agents and automated workflows using LLMs, APIs, and integration tools. Own the end to end lifecycle from prototype through production deployment, including testing, monitoring, and iteration.

• Operational Efficiency Improvement: Reduce manual effort in repetitive operational workflows by embedding AI agents into daily processes. Measure time saved, error reduction, and throughput improvements. Ensure agents maintain accuracy and auditability standards required by operations.

• Adoption and Change Management: Drive bottom up adoption by working directly with operations teams and individual contributors. Provide hands on training, create playbooks, run workshops, and build internal champions. Track adoption metrics and iterate based on user feedback

• Data Quality and Integration: Ensure AI agents operate on clean, reliable data by working with data teams and system owners. Define data requirements, build integration points with operational systems, ERP, and document management platforms, and establish feedback loops to improve agent accuracy over time.

• Performance Measurement and Reporting: Define and track KPIs for each deployed agent including adoption rates, processing accuracy, throughput gains, exception rates, and ROI. Report outcomes to operations leadership and use evidence to inform scaling decisions and investment cases.

• Cross Functional Collaboration: Partner with Product, Engineering, and IT to align agent deployments with platform capabilities, security requirements, and data governance standards. Ensure agents are built within approved frameworks and comply with company policies and regulatory requirements.

• Experimentation and Innovation: Stay current with emerging AI capabilities, LLM developments, and automation tools relevant to operations. Run structured experiments to test new approaches and share learnings across the organisation to build collective AI fluency.

• Documentation and Knowledge Management: Maintain clear documentation for all deployed agents including design rationale, configuration, dependencies, known limitation,, and escalation paths. Build a reusable library of agent patterns and templates for operational use cases.

• Vendor and Tool Evaluation: Evaluate and recommend AI tools, platforms, and vendors relevant to operational use cases. Provide informed build versus buy recommendations within group guardrails.

Skills and Knowledge

Strong understanding of LLMs, prompt engineering, and AI agent frameworks. Practical experience with automation platforms, document processing tools, and API integrations. Deep knowledge of operational workflows including document handling, data validation, exception management, and process orchestration. Data analysis and KPI design skills. Change management ability to drive adoption bottom up without formal authority. Strong communication skills to translate technical concepts for operational audiences.

Claude Enterprise and Cowork Skills

Proficiency in Anthropic s Claude platform including Claude for Enterprise and Claude Teams administration, user provisioning, workspace configuration, and usage policy management.

Hands on experience with Claude Cowork mode for desktop automation, including skill creation and management (SKILL.md authoring), custom plugin development and deployment, and scheduled task configuration using cron based scheduling.

Experience with Claude Code CLI for agentic coding workflows, including hooks, slash commands, and MCP (Model Context Protocol) server integration for connecting Claude to enterprise data sources and third party tools.

Ability to design and implement agentic loops and multi step agent workflows, including tool use patterns, iterative reasoning chains, sub agent orchestration, error recovery loops, and human in the loop approval gates.

Advanced prompt engineering for Claude models including system prompts, structured output formatting, tool definitions, chain of thought techniques, and context window optimisation for enterprise scale document processing.

Working knowledge of the Anthropic API or OpenAI including Claude Agent SDK, tool use (function calling), streaming responses, batch processing, and model selection across Claude Opus, Sonnet, and Haiku variants for cost and latency optimisation.

Any object oriented and functional language expertise including surrounding versioning and infrastructure needs.

Familiarity with enterprise connectors and integrations within Claude ecosystem including Microsoft 365 (Outlook, Teams, SharePoint), browser automation via Claude in Chrome, and third party MCP connectors for operational data sources.

Understanding of AI safety, security, and governance principles including prompt injection defence, content filtering, data privacy controls, audit logging, and compliance with enterprise security policies when deploying AI agents at scale.

Continuous learning in AI, LLMs, and automation expected.