AI Platform Lead

Descripción del puesto: Mission Design, build and evolve ESTEVE’s enterprise AI platform and agent ecosystem, establishing the reusable technical foundations, engineering standards and production controls that enable Product Teams to deliver secure, scalable and maintainable AI solutions. As AI adoption moves from pilots to production, ESTEVE needs consistent technical foundations and decision-making across business domains. Without this role, platform capabilities, architectural choices and engineering practices risk becoming fragmented, slowing reuse, increasing dependency on external partners and making secure enterprise-scale adoption harder to sustain. Key responsibilities This role provides technical leadership across the enterprise AI ecosystem but does not own individual business products or replace the accountability of Product Teams. It sets common architecture, platform capabilities and engineering guardrails, while Product Teams remain accountable for business outcomes and solution delivery. A. Build the AI platform foundations - Design and build the initial enterprise agent runtime and reusable platform components. - Implement reference architectures, reusable Agent Skills, integration patterns and deployment templates. - Create reference implementations that guide Product Teams in developing and deploying AI solutions. - Establish foundations for identity, permissions, tool access, grounding, observability and human approval. B. Industrialize AI delivery - Define and implement the standard path from experimentation to production. - Establish CI/CD, automated testing, AI evaluation, versioning, release and rollback practices. - Define minimum production-readiness criteria and verify compliance before deployment. - Ensure that solutions include monitoring for quality, reliability, latency, cost and security. C. Govern technical architecture - Maintain the target architecture, reference patterns and technology roadmap. - Lead technical design reviews for the most relevant AI initiatives. - Define approved, conditional and non-recommended technology patterns. - Manage architectural exceptions and document the associated decisions and risks. D. Enable teams and manage partners - Support Product Teams in adopting the platform and reusable components. - Provide technical guidance, examples and engineering documentation. - Review partner architectures, code and technical deliverables. - Ensure that critical knowledge, repositories and architectural decisions remain under ESTEVE’s control. Required experience - Typically, 4 to 6 years of relevant experience across AI engineering, solution architecture, digital platforms or enterprise integration, with increasing technical ownership. -  At least 2 to 4 years of hands-on experience designing and implementing generative AI, agent-based solutions, copilots, retrieval-augmented generation or enterprise AI platforms, including production-grade agent harnesses: the runtime and control layer that manages agent permissions, tool access, approvals, isolation, failure handling, recovery and reusable skills or tools. - Experience taking AI solutions beyond proof of concept into production or enterprise-scale implementation. - Experience working with business stakeholders, Product Teams, architects, cybersecurity specialists and external technology partners. - Ability to define architecture and technical direction even when implementation is delivered with external partners Technical capabilities AI & agent systems - Generative AI and agentic architecture - Agent orchestration, context and memory management, tool execution and lifecycle management - Production-grade agent harnesses, including permissions, approvals, isolation, failure handling and recovery - Design, packaging, versioning, testing and governance of reusable Agent Skills and tools - Retrieval-augmented generation and knowledge grounding - AI evaluation, observability and production monitoring - Human-in-the-loop and approval patterns - Enterprise engineering - Python or C#/.NET, REST APIs, OpenAPI, Git and automated testing - Agent and enterprise integration using MCP, function calling, API management and event-driven patterns - Cloud deployment, CI/CD, containers and infrastructure as code - Identity, managed identities, RBAC, secrets management, private networking and secure tool access - Vendor and technology assessment Preferred technology exposure Relevant hands-on exposure may include application development with Python or C#/.NET; AI platforms and frameworks such as Microsoft Foundry, Azure OpenAI, Foundry Agent Service, Microsoft Agent Framework, Semantic Kernel or LangGraph; retrieval with Azure AI Search or equivalent vector-based technologies; integration through MCP, OpenAPI, API Management, Azure Functions or Logic Apps; deployment using GitHub Actions or Azure DevOps, Docker, Kubernetes/AKS and infrastructure as code; security with Microsoft Entra ID, managed identities, Key Vault and RBAC; and observability and evaluation with Open Telemetry, Application Insights, Azure Monitor or equivalent tools. Comparable experience in other enterprise technology ecosystems is also relevant. Profile we are looking for - A practical AI technologist who can move comfortably between architecture, engineering and enterprise reality. - A hands-on builder who can create the initial agent runtimes, harnesses, reusable skills and technical patterns, then enable Product Teams and partners to adopt them consistently at enterprise scale. - A strong communicator who can explain complex technical choices in plain language and influence without formal authority. - Curious and current on market evolution, but disciplined enough to separate sustainable enterprise capabilities from short-lived trends. What success looks like - A clear and pragmatic AI architecture and technology roadmap is established. - Product Teams can reuse governed platform components, agent harnesses and Agent Skills instead of rebuilding foundations for every use case. - AI agents and solutions move from pilot to production through consistent engineering standards, automated evaluation and release controls. - AI solutions are observable, supportable and aligned with enterprise security and governance requirements. - ESTEVE is less dependent on external providers for core AI architecture decisions. - Internal teams can design, evaluate and operate AI solutions with greater technical autonomy while following common enterprise standards.