Artificial Intelligence Engineer

Client: Agency of the United Nations

Start date: ASAP.

Location: remote Europe, Valencia time zone (CET).

Full time project from Monday to Friday.

Duration: initially 3 months with possibility of extension.

Rate: Based on experience and budget availability

We're Hiring: Two AI Consultancy Roles

We're looking to fill two separate AI consultancy engagements. Please note in your application which role you're applying for — Role 1: AI for Finance or Role 2: AI Developer. You may apply for one or both.

ROLE 1 — AI FOR FINANCE (Consultant)

About the role

A specialist consultancy to operationalise a set of AI use cases for a finance function. The consultant will develop and validate the AI content these use cases depend on, enable a network of internal trainers to deliver the associated training independently, build the required solutions and automated workflows, and provide technical support to key staff — including standby technical support for a Finance AI workshop.

Context

A set of finance use cases has been identified across the offices and units concerned. Each describes an existing manual finance process, the outcome sought, and the data the solution would rely on. Between them, the use cases span reconciliation and verification, quality assurance of financial and donor reporting, budgeting and cost analysis, data analysis and reporting, controls and anomaly detection, workflow automation, visualisation and dashboards, research and due diligence, staff enablement, and document drafting and editing.

The list of use cases is a living document and is expected to grow. The consultant is expected to confirm the delivery route for each use case at inception and as work proceeds, to recommend the simplest approach that meets the requirement, and to accommodate reasonable additions and reprioritisation within the agreed scope.

Areas of work

A. Content development and validation — Convert the identified use cases into working, tested content. Begin with a proof of concept on a small number of representative use cases, demonstrated end to end and agreed before wider build. Draft and refine prompts and structured projects; assemble the reference material each solution depends on (checklists, guidance, approved templates); validate every asset before release, including by deliberately introducing known errors into test files and confirming they are detected; and record each asset in a controlled library with its version, owner, date last tested and known limitations.

B. Building agents and automated workflows — Build the required agents together with the structured projects and automated workflows that go with them. Design each agent's task, instructions, permitted data sources and tools, and its handover point back to a person; design the triggers, routing, folder structures and status tracking; define a human review and approval point in every workflow; test with genuine working data; and obtain sign-off from the owner of each solution.

C. Prompt library — Establish and maintain a controlled, version-managed prompt library covering the validated content, in place and populated before the training of trainers begins, and updated through the remainder of the engagement.

D. Training of trainers — Build internal training capacity rather than deliver end-user training directly. Design a train-the-trainer curriculum with a facilitator guide, session plans, exercises based on genuine but de-identified working files, model answers and troubleshooting notes; deliver the sessions and supervised practice runs; assess each trainer against an agreed readiness checklist; and coach trainers through their first independent deliveries. All materials to be provided in editable form.

E. Standby technical support for the Finance AI workshop — Act as the designated technical resource for a three-day Finance AI workshop. In advance, prepare and test the demonstration environment, sample data, accounts, access and equipment, and rehearse every live demonstration. During the workshop, remain on standby to resolve technical issues, support trainers and facilitators, staff a help desk, and hold contingency arrangements. Afterwards, record all issues raised, resolve those outstanding, and consolidate feedback into an improvement note.

F. Technical support and troubleshooting — Provide responsive support to key staff throughout the assignment. Diagnose and resolve issues with prompts, agents, application configuration, automated workflows, data extracts and outputs; advise on handling very large data extracts; hold regular office hours or clinics during roll-out; escalate platform, connection and licensing matters clearly; and maintain a log of issues, root causes and resolutions.

G. Documentation and handover — Produce technical and user documentation for every asset, confirm a named owner for each, and deliver a structured handover so the content and solutions can be operated, updated and extended without further external assistance.

Technical requirements

The consultant will work within the existing technology environment; no additional platform is to be introduced without prior agreement.

• Claude (Anthropic) — Primary AI platform for the assignment. Expert-level, demonstrated experience designing, testing and deploying prompts, structured projects and agents in a professional setting, including agentic workflows, analysis of documents and spreadsheets, and disciplined validation and version management of content.
• Generative AI — Substantive experience applying generative AI to business processes beyond a single tool: prompt design and evaluation, grounding responses in source documents, handling model limitations such as inconsistency and fabricated content, and judgement on where generative AI is and is not appropriate. Familiarity with the wider market and how the field is developing.
• Microsoft 365 and Power Platform — Excel, Word, Outlook, PowerPoint, SharePoint and Teams as the principal working environment; expert command required, including large and complex workbooks. Demonstrated build experience in Power Automate for trigger-based workflow, routing, notification and status tracking; working knowledge of Power BI and Power Query. Awareness of Microsoft Copilot.
• Enterprise financial data (e.g. Oracle Cloud Fusion) — Working knowledge of extracting and interpreting financial data through standard reports and data extracts. Solutions needing a direct system connection are designed first and built once such a connection is approved.
• SQL and relational data sources — Extraction, preparation and reconciliation of high-volume transaction populations spanning multiple projects, offices and periods.
• Finance process knowledge — Practical understanding of reconciliation, period-end closure, donor and grant reporting, budget preparation and review, and internal control.
• Training of trainers — Proven record of designing and delivering train-the-trainer programmes and enablement materials for non-technical staff.
• Technical support — Hands-on user and event support for business applications and automations, including standby support at live events, with disciplined issue logging and escalation.
• Adaptability — Ability to work from an evolving requirement and to advise candidly when a simpler solution will serve better.
• Data protection and governance — Understanding of data confidentiality, records management and responsible-use practice for AI, particularly around personal, financial and third-party information.

ROLE 2 — AI DEVELOPER

About the role

Low-code AI development and technical delivery of an AI solution, covering agent build, retrieval grounding, WhatsApp channel delivery, security assurance and handover, in support of a multilingual public-information platform and its AI-powered chatbot.

The engagement runs for four months and is remote. The consultant starts against an approved architecture and a signed-off requirements baseline, so all four months are build, assurance and handover rather than discovery. The work draws on the Microsoft AI ecosystem, including Copilot Studio, Azure AI Foundry and the Power Platform.

Education, experience and skills required

• Master's degree in Computer Science or a related field from an accredited institution with two years of relevant professional experience; or a university degree in the above fields with four years of relevant professional experience.
• Hands-on experience developing and deploying AI agents using low-code/no-code tools such as Microsoft Copilot Studio, Power Apps and Azure AI Studio, with a focus on real-world business use cases.
• Proven experience integrating AI capabilities into business workflows using the Microsoft Power Platform (Power Automate, Power Apps, Power Virtual Agents) and Azure AI services, including Azure OpenAI and Cognitive Services.
• Strong practical knowledge of Microsoft Azure, particularly services supporting AI development, orchestration and data access — Azure AI Foundry, Azure AI Search, Azure Functions, Logic Apps and Azure Data Lake.
• Demonstrated experience implementing retrieval-augmented generation and grounding patterns, including indexing, chunking and retrieval tuning.
• Experience designing and operating CI/CD pipelines and applying security scanning, remediation and access-control measures within Azure environments.
• Demonstrated ability to troubleshoot and maintain low-code AI solutions, including multi-agent or coordinated-agent systems in enterprise environments.
• Experience collaborating with business users and external partners to co-design, test and refine AI agents and automation tools.
• Background in software or application development, with familiarity in Power Platform development, prompt engineering and citizen-developer enablement a distinct advantage.
• Proficiency in English required. Knowledge of French is an asset, given the platform's bilingual user base.