Principal AI Solution Architect

KData AI; United States; Remote; CAD 83,000 - 165,000; Part time

Job summary

KData AI is seeking a Principal AI Solution Architect for a remote, senior contractor role. You will lead hands-on architecture across GenAI, RAG, agentic AI and related platforms, delivering scalable reference implementations and end-to-end use cases on a reusable foundation.

Partner with engineering, data and control teams to ensure secure, production-ready solutions in Azure Foundry, Databricks and open-source tech.

Qualifications

• Must own production GenAI, RAG or intelligent automation architectures.
• Proficiency in Python and API development for production-grade code.
• Ability to validate architecture using Azure and Databricks environments and traces.
• Experience in Microsoft AI ecosystem and Foundry implementations.
• Experience delivering end-to-end GenAI platform patterns and use cases.
• Broad depth across application, integration, data, cloud, security and operations.

Responsibilities

• Own target-state and transition-state architectures from discovery to production.
• Design and validate reusable RAG, agentic and document-processing patterns.
• Define platform boundaries for model access, agents, tools, and registries.
• Lead technology decisions across Microsoft Foundry, Azure AI and Databricks.
• Define model/endpoint abstraction, routing, quotas, and provider selection.
• Ensure identity, least-privilege, data isolation and secure integrations.
• Set non-functional expectations for quality, availability, and cost.
• Embed security, privacy and Responsible AI controls throughout.
• Progress PoC to MVP and production across data, environments, and controls.
• Maintain architecture decision records and risk assessments.
• Review platform code and provide actionable feedback.

Skills

GenAI architecture ownership

Python & API development

Azure & Databricks validation

Microsoft AI ecosystem knowledge

End-to-end GenAI delivery

Cross-domain knowledge (app, data, sec

Tools

Azure

Databricks

MLflow

Unity Catalog

Azure OpenAI

Job description

This is a remote position.

Principal AI Solution Architect

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

Role Overview

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.

Key Responsibilities

• 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 organization, integration contracts, deployment patterns and non-functional evidence for conformance, modularity, security, reliability, operational readiness and appropriate reuse. Provide actionable code-level feedback and verify resolution.

Requirements

Required Qualifications

• Substantial architecture ownership for production GenAI, RAG, agentic AI or intelligent automation in complex enterprises.
• Current Python and API-development proficiency to build, debug and review production-quality code; recent shared AI platform/application repository contributions through reference implementations, reviews, automated tests or production troubleshooting.
• Ability to validate architecture using Azure and Databricks environments, deployment configuration, traces, evaluations and working code.
• Current practical Microsoft AI ecosystem and Microsoft Foundry knowledge supported by implementation or technical evaluation, rather than certification alone.
• Experience implementing a production or production-relevant GenAI platform pattern and delivering an end-to-end use case.
• Depth across application, integration, data, cloud, security and operations; strong knowledge of agents, tools, MCP, retrieval, APIs, distributed systems, identity, observability, evaluation and production governance.
• Experience defining reusable platforms, reference architectures, standards and experimentation-to-production migration paths, with clear trade-off explanations, decision documentation and actionable guidance.

Preferred Qualifications

• Financial services, regulated-industry or sensitive-information experience.
• MLflow, Unity Catalog, Azure AI Search, containers, event-driven integration, API management and infrastructure as code.
• Build-versus-buy assessments and approved open-source agent/orchestration frameworks.
• Multimodal, document-intelligence or human-review solution design.

Success and Role Boundaries

Deliver coherent architectures that progress to production without fundamental redesign, reduce duplicated engineering, verify controls before release and provide tested reference implementations. Prove the foundation through at least one end-to-end use case, improve code quality and enable internal engineers to operate and extend shared capabilities.

The Lead Full-Stack GenAI Developer owns most use-case implementation, full-stack integration, delivery automation and production troubleshooting. Internal owners retain business prioritization, backlog and acceptance; enterprise, data, infrastructure and control functions retain their mandates and formal approvals.

The Architect does not replace formal enterprise architecture or control approvals. Security, privacy, risk and Responsible AI functions retain independent review or approval responsibilities.

Candidates who no longer actively write, debug and review code—including diagram-only architects, governance coordinators, delivery managers, pre-sales architects or vendor-product specialists—do not meet the requirement.