GenAI Expert

Mission Objective

Under the supervision of the AI System Architect, the main role is to design, implement, and evolve Gen AI assets on a hybrid GCP/Microsoft platform, with a focus on scalability and performance.

The candidate must ensure the technical consistency, scalability, and robustness of Generative AI solutions and act as a reference expert for LLM architectures.

Tasks Include

• Gen AI Architecture: Designing decoupled, scalable, and secure architectures (RAG, autonomous agents), defining application architecture for LLM environments (DEV, UAT, PROD), and integrating CI/CD through LLMOps.
• Augmented Developer: Proven experience with “Augmented Developer” solutions is required to successfully handle part of the role’s responsibilities.
• Governance & Technical Debt: Implementing AI governance principles and managing technical debt.
• Robust Architectures: Designing resilient and fault-tolerant architectures, writing interface contracts, developing fallback processes, and managing non-functional requirements (latency, security, reliability).
• Interaction & Ingestion Models: Defining optimized models, analyzing business needs, creating conceptual models for vector storage, optimizing LLM token usage, collaborating with development teams, and implementing continuous evaluation mechanisms.

Expected Technical Skills

• GCP: Agent Platform (formerly Vertex AI), BigQuery Data Agents
• Microsoft: Copilot, Copilot Studio, WorkIQ, M365 Copilot Agents
• Programming: Python, SQL
• Frameworks & Protocols: ADK, A2A, MCP

Specific Expertise & Knowledge Expected in GCP, LLMOps & Cognitive Architectures

• Orchestration & Function Calling Expertise: Proven experience in decoupling business logic to dynamically interact with Information Systems, ensuring modular and reusable code (Python, Go, etc.).
• GCP & Microsoft AI Ecosystems: Deep knowledge of Agent Platform (formerly Vertex AI), Gemini, and equivalent Microsoft platforms. Ability to select the appropriate models based on cost/performance/complexity trade-offs and work within a multi-model cloud environment.
• Advanced Design for High Volumes & Complex Use Cases:
  • Large-scale vector database management (Vector Search)
  • Advanced semantic caching strategies to reduce latency and costs
  • Mastery of agent-based architectures (multi-agent systems, planning, reasoning)
• Gen AI Security Requirements:
  • Knowledge of prevention strategies against Prompt Injection and Data Poisoning attacks
  • Implementation of strict filtering mechanisms (semantic OLS/RLS) to ensure LLMs only return data accessible to the user
• Observability & Monitoring (LLMOps): Knowledge of best practices for monitoring AI application chains (prompt traceability, response times, token costs) and forwarding logs to enterprise security layers (SIEM).