AI Support and Integration Engineer

Job description

Role purpose The AI Support and Integration Engineer sits at the intersection of client services and applied AI. The role owns client tickets and requests that involve AI tooling — from day-to-day usage questions on Claude Desktop and Microsoft 365 Copilot through to building integrations against the Anthropic API, Microsoft Graph, Power Automate, and Azure Function Apps. This person is the go-to for clients adopting AI in their workflows, and the internal specialist who keeps the team current on a fast-moving toolchain. Clients are primarily hedge funds and financial services businesses based in London and New York, which is why this role runs a split shift covering both UK and US Eastern business hours. Success means clients get fast, accurate resolution of AI-related tickets; new integrations ship reliably; and the business stays ahead of the curve as models, agent frameworks, and productivity tools evolve. Working pattern This is a rostered split-shift role. You will work a combination of two shifts across the month: • UK shift — 10:00 to 19:00 SAST. Covers the UK working day end-to-end. • US Eastern shift — 15:00 to 00:00 SAST. Covers the US Eastern working day end-to-end. Attracts a +15% shift allowance on hours worked, and catered dinner will be provided in-office. During your probation period you will work in our Cape Town office. After successful probation the role moves to hybrid, minimum 3 days per week in-office. Cover for both shift patterns is expected from the start — the in-office period is about onboarding into the team, the tools, and the client base. Key responsibilities Client support and ticket resolution • Own AI-related client tickets end to end — triage, diagnose, reproduce, resolve, and document. Typical tickets span Claude Desktop and Claude Code configuration, Cowork setup, Copilot licensing and agent behaviour, Office add-in issues, Power Automate flow failures, and API integration problems. • Act as the escalation point for the wider support team on anything AI-related. Coach first-line colleagues so recurring issues get handled at tier 1. • Maintain runbooks and knowledge base articles for common AI tooling scenarios — setup, troubleshooting, known limitations, and workarounds. • Communicate clearly with non-technical stakeholders about what AI tools can and cannot do, managing expectations around model behaviour, hallucinations, and data handling. Integration and solution delivery • Build and maintain integrations between AI platforms and client systems — Anthropic API, OpenAI, Google Gemini, Microsoft Graph, M365 services (SharePoint, Teams, Outlook, Exchange), and line-of-business apps. • Deliver automation using Power Automate and Azure Function Apps — designing flows, triggers, connectors, and serverless components that plug AI into existing client workflows. • Develop Office add-ins (Word, Excel, Outlook) that embed AI features into day-to-day workflows, including authentication, manifest management, and deployment via the Microsoft 365 admin centre. • Configure and deploy Copilot agents and Copilot Studio solutions for clients, including grounding on SharePoint and Dataverse, prompt design, and governance controls. • Assist clients with Claude Code, Claude Desktop, and Cowork rollouts — tool setup, MCP server configuration, skills, and integration with existing dev workflows. • Write clean, maintainable code — predominantly Python, TypeScript/JavaScript, and PowerShell — with appropriate error handling, logging, and secrets management. Enablement and emerging tech • Track emerging AI tooling — new model releases, agent frameworks, MCP servers, Claude Skills, and productivity integrations — and evaluate what is worth bringing into the stack. • Run internal demos and client workshops on new capabilities, translating raw product updates into practical guidance for specific client contexts. • Pilot and prototype new agent and automation patterns, scoping small proofs of concept before wider rollout. • Contribute to AI governance and acceptable-use guidance covering data residency, PII handling, prompt-injection risks, and model selection. Required skills and experience Systems engineering foundation • 5+ years of systems engineering or integration experience in a client-facing or MSP environment. • Microsoft 365 depth — Entra ID, Graph API, Exchange Online, SharePoint, Teams administration. • Power Automate — flow design, triggers, standard and custom connectors, error handling, and deployment across tenants. • Azure Function Apps — authoring, deployment, scaling, and integration with Logic Apps and Graph. • PowerShell fluent for M365 administration and automation. Claude platform • Hands-on experience with Claude Desktop as a daily-driver — MCP servers, Skills, Projects, and file handling. • Working knowledge of Claude Code for agentic coding tasks, including custom slash commands, sub-agents, and hooks. • Exposure to Cowork for non-developer automation of files and tasks. • Practical experience calling the Anthropic Messages API, including tool use, streaming, batching, and prompt caching. Microsoft 365 and Copilot • Microsoft 365 Copilot — licensing models, admin configuration, and real-world use across Word, Excel, Outlook, Teams, and PowerPoint. • Copilot Studio — building declarative and custom agents, topics, knowledge sources, and actions. • Office add-in development using the Office JS API and manifest deployment through the Microsoft 365 admin centre. • Azure fundamentals relevant to AI — Azure OpenAI, Logic Apps, Key Vault, and related services. Other AI platforms • Google Gemini — API usage, Gemini in Workspace, and comparative positioning versus Claude and Copilot. • Awareness of other models and agent frameworks in active use — OpenAI GPT, Mistral, and open-weight models where relevant. • Familiarity with MCP (Model Context Protocol) and agent patterns across vendors. Engineering fundamentals • Strong scripting and development skills in Python and TypeScript/JavaScript. • REST API integration, OAuth 2.0, webhooks, and event-driven patterns. • Git-based workflows, CI/CD basics, and sensible secrets management. • Debugging across client-side, server-side, and third-party API boundaries. Ways of working • Ticket-driven support experience — able to triage, prioritise, and communicate progress clearly. • Writes clean, concise documentation aimed at both technical and non-technical readers. • Comfortable working directly with clients, including running scoping calls and demos. • Pragmatic, security-aware, and careful with client data. • Able to sustain a rotating split-shift pattern with the discipline that requires. Nice to have • Microsoft Foundry (AI Foundry) — agent building, model deployment, and evaluation workflows. • Microsoft Fabric — OneLake, data integration, and Copilot grounding on Fabric data. • SharePoint Framework (SPFx) and wider Power Platform (Power Apps, Dataverse). • Prior exposure to financial services clients — hedge funds, asset managers, or PE — and associated compliance controls. • Certifications such as MS-102, AI-102, AZ-104, or equivalent. • Open-source contributions to MCP servers, Claude Skills, or agent tooling. What we offer • Competitive senior salary plus a 15% shift allowance on US Eastern hours worked and catered dinner provided in-office. • Hybrid working after probation. • A front-row seat on the most interesting platform shift in years, working across Anthropic, Microsoft, and Google stacks. • Direct client impact — your work shows up in how real teams get their jobs done. • A team that takes AI seriously but does not hype it. How to apply Send a CV and a short note describing a recent AI-related problem you solved — what the ticket or request looked like, what you did, and what you would do differently next time. Links to working integrations, Skills, MCP servers, or demos are welcome in place of a long write-up. Please also confirm you are able to work the split-shift pattern described above.