AI Solutions Architect

Job Summary The AI Solutions Architect is the senior technical leader for KLS Martin's enterprise AI program and the technical partner to IT leadership and the AI Program Manager. This role designs and builds the foundational AI platforms, services, and reference architectures that power AI adoption across the company, sets the technical standards that federated teams build to, and ensures every AI implementation meets enterprise requirements for security, scalability, reliability, and regulatory compliance.

This is a hands-on architect role: the person in this seat personally designs and ships the foundational systems and the highest leverage production use cases and grows technical leadership responsibilities as the AI engineering team expands.

The AI Solutions Architect partners with stakeholders across IT, business units, Quality, Regulatory, Information Security, and the AI Program Manager to ensure AI solutions deliver on their objectives while meeting applicable regulatory, quality, and security standards.

The AI Solutions Architect contributes to the continuous improvement of AI development practices, reference architectures, evaluation frameworks, and developer enablement, and develops the capability of the AI engineering team through coaching, mentoring, and targeted assignments as the team grows.

Essential Functions, Duties, and Responsibilities Personally design, build, and ship foundational AI systems, platforms, and integrations, including agent frameworks, evaluation infrastructure, and the first wave of production use cases.

Lead end to end implementation of the most strategic AI initiatives across the company.

Maintain hands on proficiency with the production stack:
Python, FastAPI, React, Azure services, and modern AI tooling.

Define, document, and evolve KLS Martin's reference architectures for the most common AI patterns: retrieval augmented generation, autonomous agents, document processing, classification, and assistive copilots.

Build and curate the internal prompt and pattern library so federated teams reuse rather than rebuild.

Establish coding standards, security controls, evaluation requirements, and documentation expectations for AI development across the enterprise.

Contribute to the AI technology roadmap in partnership with IT leadership and the AI Program Manager, including model providers, platforms, frameworks, and supporting tooling.

Lead build versus buy decisions for AI capabilities, including technical evaluation of vendor solutions against internal builds

Provide the technical backbone for vendor selection, proof of concept work, and contract negotiation.

Design and operate the MLOps and LLMOps stack, including deployment pipelines, model and prompt versioning, evaluation harnesses, monitoring, and observability.

Build evaluation and testing frameworks, including eval datasets, regression testing, and red team testing for prompt injection, jailbreaks, and data exfiltration.

Implement cost and token observability for API spend management.

Lead architectural and technical response to AI specific incidents such as production hallucinations, data leakage, prompt injection, and model availability issues.

Design and implement identity, secrets, and access management patterns for AI systems, including agent identities, least privilege patterns, and integration with Microsoft Entra e data classification, retention, and access controls are designed into every solution from the start.

Partner with Information Security on threat modeling and risk reviews for new AI systems.

Ensure AI architectures meet validation requirements when use cases touch quality processes or regulated data under ISO 13485 and FDA 21 CFR Part 11.Provide architecture review, code review, and technical mentoring for federated developers and citizen builders working on AI features in their own areas.

Lead, mentor, and develop AI engineers added to the team over time, including direct people management responsibilities as the team grows.

Maintain a graduation process for moving prototypes from POC into supported production systems, including handoff criteria, runbooks, and on call expectations.

Represent AI…