AI Architect
AI Architect for a 12-month on-site consultancy mission at the Federal Pension Service in Brussels: define target and reference architectures for generative, agentic and ML AI (gateways, MCP, retrieval, model serving, observability) and translate security/privacy rules into architectural guardrails. Dutch and French required; part-time telework possible.
- Compensation
- Not disclosed
- Location
- Bruxelles, Belgium
- Work setup
- Not specified
- Seniority
- Not specified
- Employment
- Part time
- Experience
- Not specified
- Category
- Architecture
- Posted
Skills
Must have
- Extensive experience in solution architecture or IT architecture in enterprise environments
- Experience with API, event, and integration patterns/architectures
- Understanding of LLM and AI platform architectures: retrieval, model serving, agents, evaluation, observability
- Ability to translate regulations and security requirements into architectural principles and guardrails
- Experience with architectural decision records, reference architectures, non-functional requirements, and lifecycle management
- Dutch — active knowledge
- French — active knowledge
- English — passive knowledge
- 2-page application document outlining a vision for safe, responsible AI at the Federal Pension Service
Nice to have
- Knowledge of the public sector, G-Cloud, or OISZ context
Full description
Mission Overview:
We are seeking an AI Architect for a consultancy mission at a client site, representing Keystone Solutions. The AI Architect will define and oversee the target architecture, reference architectures, and integration patterns for AI. This role ensures that models, agents, data flows, gateways, platform services, and applications coherently align with the broader architecture and meet security, privacy, continuity, and manageability requirements.
Responsibilities:
- Develop and maintain the target architecture for generative AI, machine learning, and agentic AI.
- Establish reference architectures and decision patterns for on-premises, cloud, and hybrid deployments, tailored to sensitivity, integration needs, and operational requirements.
- Define architectural frameworks for AI Gateway, Model Context Protocol, retrieval, model access, agents, identity, logging, observability, and lifecycle management.
- Guide delivery teams in architectural choices, trade-offs, and the transition from sandbox or PoC to production in collaboration with technology architects.
- Conduct architecture reviews and timely highlight deviations, technical debt, dependencies, and risks.
- Integrate privacy, least privilege, data minimization, isolation, and traceability by design in collaboration with security and data functions.
- Identify and prioritize reusable platform and integration building blocks with the AI Enabling Team and the Internal Developer Platform.
- Translate technological evolutions into concrete impacts, options, and decision points for the client.
Technical Competencies:
- Extensive experience in solution architecture or IT architecture in enterprise environments.
- Experience with APIs, event, and integration patterns.
- Understanding of LLM and AI platform architectures, retrieval, model serving, agents, evaluation, and observability.
- Ability to translate regulations and security requirements into practical architectural principles and guardrails.
- Experience with architectural decision records, reference architectures, non-functional requirements, and lifecycle management.
- Knowledge of the public sector, G-Cloud, or OISZ context is a plus.
Generic Competencies:
- Innovation: You keep up with the latest trends and experiment with innovative AI technologies.
- Problem-solving: You analyze complex issues and translate them into pragmatic solutions.
- Collaboration: You work smoothly with colleagues from various disciplines (IT, business, security, architecture).
- Communication: You can clearly explain technical concepts to both technical and non-technical stakeholders.
- Advising: You excel in structuring, visualizing, advising, and facilitating multidisciplinary decision-making.