Job Description: Generative AI Engineer
1) Solution Design & Development
- Design, build, and deploy custom copilots and Copilot extensions (plugins, connectors, actions) using Copilot Studio and Graph connectors.
- Engineer prompt flows, retrievalaugmented generation (RAG), and grounding via SharePoint, Dataverse, Azure Cognitive Search, and Microsoft Graph.
- Develop Copilot for M365 skills across Teams, Outlook, Word, Excel, PowerPoint, and Loopoptimizing grounding, context, and guardrails.
- Create Power Automate integrations and Power Apps UI components to embed copilots into business processes.
2) Data & Integration Engineering
- Implement data connectors and Graph APIs to securely surface enterprise content with appropriate access controls (AAD, Entra ID, Graph scope).
- Build and optimize vector indexes/semantic search (Azure AI Search) for domainspecific RAG patterns.
- Integrate with LOB systems (e.g., SAP, Salesforce, ServiceNow) using connectors or custom APIs; manage schema, metadata, and content governance.
3) Safety, Security & Compliance
- Apply Responsible AI and Trustworthy AI principles: content filtering, prompt safety, redteaming, and misuse prevention.
- Enforce data residency, DLP, eDiscovery, retention labels, and Conditional Access for Copilot interactions.
- Partner with Security/GRC to ensure solutions meet internal controls (e.g., ISO 27001/NIST) and organizationspecific policies.
4) MLOps & Operations
- Establish DevOps/MLOps for Copilot assets: environments, source control, CI/CD, versioning, and telemetry.
- Monitor quality, latency, tokens/costs, usage analytics, and feedback loops to iteratively improve prompts and RAG performance.
- Implement A/B testing, prompt evaluations, and evaluation datasets for continuous improvement.
5) Stakeholder Enablement & Change
- Run discovery workshops, translate use cases into Copilot capability maps and ROI hypotheses.
- Produce documentation (solution designs, prompt handbooks, runbooks) and deliver enablement/training to business users.
- Advise product teams on prompt patterns, hallucination mitigation, and governance.
