Primitives

fai-llm-landscape-expert

LLM landscape expert — model families (GPT, Claude, Llama, Gemini, Phi), benchmarks (MMLU, HumanEval, MT-Bench), deployment types, quantization, and model selection frameworks.

• cost-optimization
• performance-efficiency

fai-ml-engineer

ML engineering specialist — model training pipelines, LoRA/QLoRA fine-tuning, evaluation metrics, MLOps with Azure AI Foundry, model registry, and serving optimization for production AI.

• cost-optimization
• performance-efficiency
• operational-excellence

fai-performance-profiler

Performance profiling specialist — latency analysis (P50/P95/P99), token optimization, GPU utilization profiling, bottleneck identification, cold start analysis, and AI pipeline performance tuning.

• performance-efficiency
• cost-optimization

fai-play-dispatcher

FAI play dispatcher — routes user requests to the correct solution play based on intent classification, understands all 101 plays and their domains, delegates to specialist agents.

• cost-optimization
• operational-excellence

fai-power-platform-expert

Power Platform specialist — Power Apps (Canvas + Model-driven), Power Automate cloud flows, Dataverse, custom connectors, DLP policies, and Copilot Studio integration.

• operational-excellence
• cost-optimization

fai-product-manager

AI product management specialist — requirements gathering, AI use case prioritization, evaluation metric design, go-to-market strategy, and responsible AI impact assessment.

• responsible-ai
• cost-optimization
• operational-excellence

fai-production-patterns-expert

Production AI patterns expert — hosting selection (Container Apps/AKS/Functions), APIM gateway patterns, streaming SSE, retry/circuit-breaker, health checks, and multi-region deployment for production AI workloads.

• reliability
• performance-efficiency
• cost-optimization
• operational-excellence

fai-prompt-engineer

Prompt engineering specialist — system message design, few-shot patterns, chain-of-thought, structured output schemas, anti-hallucination techniques, and evaluation-driven prompt optimization.

• reliability
• responsible-ai
• cost-optimization

fai-rag-architect

Enterprise RAG architecture specialist — designs end-to-end retrieval-augmented generation pipelines with Azure AI Search, OpenAI embeddings, chunking strategies, grounding, citation, evaluation, and production deployment.

• security
• reliability
• cost-optimization
• performance-efficiency

fai-rag-expert

RAG expert — advanced retrieval patterns (agentic, graph, multi-modal RAG), chunking strategies, hybrid search, re-ranking, evaluation metrics, and production RAG optimization.

• reliability
• performance-efficiency
• cost-optimization

fai-ray-expert

Ray distributed computing specialist — Ray Serve for model serving, Ray Tune for hyperparameter optimization, Ray Data for preprocessing, and distributed training/inference at scale.

• performance-efficiency
• cost-optimization

fai-redis-expert

Redis specialist — caching patterns (TTL, LRU), semantic cache for AI (embedding similarity), pub/sub messaging, Redis Streams, and session storage for chat applications.

• performance-efficiency
• cost-optimization

fai-solutions-architect

Cloud solutions architect — end-to-end AI solution design, Azure service selection, multi-service integration, cost estimation, WAF trade-off analysis, and architecture decision records.

• reliability
• cost-optimization
• performance-efficiency
• security

fai-swarm-supervisor

Multi-agent swarm supervisor — routes tasks to specialist agents, manages turn limits and token budgets, handles agent coordination, conflict resolution, and synthesizes multi-perspective results.

• reliability
• cost-optimization