Principal AI Solution Architect

This is a remote position.

GenAI, Agentic AI & Platform | Senior Contractor / Consulting Role

Provide deeply hands-on architecture leadership across a portfolio of secure, reusable, production-ready GenAI, RAG, agentic AI, intelligent document-processing, multimodal and AI-enabled applications. Own architecture coherence, reusable platform direction, standards and technical assurance, connecting immediate use-case delivery with repeatable capabilities.

Work as one of two senior contractors, partnering directly with the Lead Full-Stack GenAI Developer and internal engineering, data, infrastructure, architecture and control teams. Build technical spikes and deployable reference implementations, contribute shared-platform code, review implementation and materially co-implement at least one representative end-to-end use case on the reusable foundation.

Core ecosystem:
Microsoft Azure, Microsoft Foundry, Databricks and approved open-source technologies.

• Own target-state, transition-state and use-case architectures from discovery through production, covering applications, agents, tools, models, enterprise data, APIs, user channels, workflows and operational controls.
• Design, implement and validate reusable RAG, agentic RAG, document-processing, multimodal, tool-calling, deterministic workflow-plus-agent and human-in-the-loop patterns.
• Define platform boundaries and co-build model access, agent execution, tools, gateways, registries, retrieval, evaluation, tracing, observability, feedback, release controls and GenAIOps foundations.
• Lead evidence-based technology decisions across Microsoft Foundry, Azure AI capabilities, Azure OpenAI, Azure AI Search, Databricks, MLflow, Unity Catalog, Azure application/integration services and approved open-source frameworks. Enable model, provider, framework, orchestration and vector-store flexibility without unnecessary complexity.
• Define model/endpoint abstraction, routing, quotas, fallback and controlled provider selection; agent, tool, prompt, API and MCP connector registration/lifecycle; and versioning of software, prompts, configuration, models, agents and evaluation assets.
• Establish identity, least-privilege authorization, separation of duties, permission-aware retrieval, data isolation and lineage, network boundaries, private connectivity, secrets, key management and secure service integration.
• Define measurable non-functional expectations for quality, availability, scalability, latency, throughput, concurrency, resilience, recovery, continuity, supportability, maintainability, capacity and cost, proportionate to each use case. Establish evaluation thresholds, release evidence, human review, monitoring, auditability, provenance and source evidence.
• Embed security, privacy, data governance, risk and Responsible AI controls, including human authorization before consequential actions, safe failure, escalation, content/tool restrictions, auditable overrides and traceability from user request through downstream actions.
• Define progression from PoC to MVP and production across data, environments, controls, testing, deployment, monitoring, support and recovery. Own architecture decision records, interface standards, technical risks, dependencies and assurance checkpoints.
• Validate decisions through working code, tests, traces, benchmarks and deployment evidence. Research and evaluate current technologies, creating working comparisons when recommendations depend on technical behaviour. Troubleshoot cross-platform issues, mentor engineers and transfer knowledge; use implementation lessons to continuously improve architecture and reusable capabilities.
• Review platform and use-case code, code…