Copilot Studio & Power Platform Engineer

Copilot Studio & Power Platform Engineer
A hands-on Microsoft Copilot Studio, Power Platform, and prompt-engineering specialist to polish, automate, and scale an existing consumer-intelligence agent. Works closely with the internal owner who built it.
The Consumer Intelligence System (CIS) is a Microsoft Copilot Studio agent, grounded on curated research and data in SharePoint / M365, that answers consumer-intelligence questions for the business. A working version has already been built for one product category, including a five-layer knowledge model, department playbooks, and a confidence-rated evidence scheme.
This role takes that build the last mile: polishing and publishing it to users in Teams, automating the manual steps that currently make it hard to maintain, and replicating it across additional categories. It is a build-and-automate role working alongside the subject-matter owner who created the system — not a ground-up platform build.
Key responsibilities
• Polish and extend the existing Copilot Studio agent, and publish it to Teams / M365 for end users.
• Fix citation carry-through: rework the consolidation prompts so every fact retains a traceable link to its original source.
• Review and optimise the master and consolidation prompts for quality, consistency, and reduced hallucination.
• Build Power Automate flows that export agent output to Word / PowerPoint and deliver it to users via download links.
• Automate data ingestion from Power BI (POS and shopper/demographic data) into SharePoint / the agent’s knowledge, replacing the current manual export step.
• Set up a repeatable, scheduled refresh of the research content on a sensible cadence, automating what can be automated across the AI research tools.
• Replicate the pattern across new categories, customising playbooks and markdown cards per category.
• Maintain the evidence / confidence scheme and the one-fact-per-layer structure, and support a lightweight quality check so review scales.
Required skills & experience
• Hands-on Microsoft Copilot Studio: building and publishing agents, configuring knowledge sources, topics, instructions, and Teams / M365 publishing.
• Strong Power Platform / Power Automate: flows, scheduled flows, SharePoint and Office connectors, and generating Word / PowerPoint output.
• Demonstrated prompt engineering: research and consolidation prompts, source grounding and citation, and controlling hallucination.
• Solid SharePoint and M365 skills: site and library structure, output folders, and permissions.
• Power BI: navigating and exporting reports, and ideally automating those exports.
• Comfort handling CSV / Excel data and consolidating third-party datasets.
• Self-directed and collaborative: able to work closely with a subject-matter owner and communicate clearly.
Preferred skills
• Familiarity with AI research tools (Copilot Researcher, Copilot Deep Research) and ideally Waldo FYI.
• Consumer / CPG or market-research domain exposure.
• Microsoft Graph or Dataverse; document generation via Office / Graph APIs.
• Awareness of content governance — evidence ratings, source traceability, and refresh cadence.
Deliverables
• A polished, published agent live in Teams, with working source citations.
• A Word / PowerPoint export feature with user-facing download links.
• Automated Power BI → SharePoint ingestion for POS and shopper data.
• A repeatable, scheduled content-refresh process.
• Replicated agents for the additional categories, with per-category playbooks and cards.
Note on fit This is a Copilot Studio + Power Platform + prompt-engineering role, not a data-engineering or data-warehouse role. The system already exists and is document-grounded; the priority is polishing, automating, and replicating it. Candidates whose strength is purely data pipelines or semantic modelling are unlikely to be the right fit weight toward Copilot Studio and automation depth.

Work arrangement
No