AI Architect – Generative & Agentic AI

The AI Architect will lead the architecture and technical direction of enterprise agentic AI initiatives for our client, driving scalable and governed AI solutions across Google Gemini, IBM Enterprise Advantage, SAP AI, Salesforce Einstein, and Databricks platforms. This role combines hands-on architecture leadership with enterprise AI governance and cross-platform orchestration expertise in a highly strategic client environment.

Your primary responsibilities will include:

• Design and architect enterprise agentic AI solutions using Google Gemini, Vertex AI, Agent Builder, and IBM Enterprise Advantage platform services.

• Define multi-agent and agent-to-agent orchestration patterns across SAP AI (BTP/Joule), Salesforce Einstein, Databricks, and enterprise systems.

• Lead architecture and implementation of long-running AI workflows including task coordination, state management, memory handling, tool use orchestration, and workflow resiliency.

• Build enterprise AI integration patterns leveraging APIs, event-driven architecture, microservices, and cloud-native services on GCP.

• Develop Retrieval-Augmented Generation (RAG) and enterprise knowledge solutions using Databricks data products and enterprise data platforms.

• Implement AI governance, guardrails, safety filters, access controls, auditability, and compliance mechanisms aligned to regulated utility industry requirements.

• Define AI solution standards, reusable architecture patterns, observability controls, and operational governance for production AI systems.

• Collaborate with client stakeholders, business teams, AI engineers, platform teams, and enterprise architects to align AI architectures with business objectives.

• Provide technical leadership and architecture governance for the Romania-based AI engineering team, including design reviews, implementation guidance, and engineering standards.

• Support AI solution estimation, roadmap definition, technical risk management, and production readiness reviews.