Primitives

fai-play-13-tuner

Fine-Tuning Workflow tuner — learning rate scheduling, LoRA rank/alpha optimization, epoch calibration, batch sizing, data quality filtering, and training cost analysis.

• cost-optimization
• performance-efficiency

fai-play-14-builder

Cost-Optimized AI Gateway builder — APIM AI gateway, semantic caching with Redis, smart model routing by complexity, token budget enforcement, and multi-region load balancing.

• cost-optimization
• performance-efficiency
• reliability

fai-play-14-reviewer

Cost-Optimized AI Gateway reviewer — routing accuracy audit, cache quality verification, budget enforcement testing, security review, and cost savings validation.

• security
• cost-optimization

fai-play-14-tuner

Cost-Optimized AI Gateway tuner — semantic cache threshold calibration, routing complexity boundaries, token budget tiers, rate limits, and cost optimization analysis.

• cost-optimization
• performance-efficiency

fai-play-15-builder

Document Processing builder — GPT-4o Vision multi-modal analysis, Azure Document Intelligence, table extraction, classification pipeline, and PII-aware storage.

• reliability
• security

fai-play-15-reviewer

Document Processing reviewer — multi-modal extraction accuracy audit, table parsing quality, PII masking verification, pipeline error handling, and classification accuracy review.

• security
• responsible-ai

fai-play-15-tuner

Document Processing tuner — GPT-4o Vision resolution settings, extraction confidence calibration, table detection thresholds, chart parsing config, and batch performance optimization.

• cost-optimization
• performance-efficiency

fai-play-16-builder

Copilot Teams Extension builder — M365 Copilot declarative agent, Microsoft Graph API integration, Adaptive Cards, Entra ID SSO, and Teams message extension development.

• security
• reliability
• operational-excellence

fai-play-16-reviewer

Copilot Teams Extension reviewer — Graph permission audit, SSO flow testing, Adaptive Card rendering review, message extension validation, and Teams policy compliance.

• security
• responsible-ai

fai-play-16-tuner

Copilot Teams Extension tuner — Graph scope minimization, knowledge source config, response tone calibration, Adaptive Card optimization, and permission tuning.

• security
• performance-efficiency

fai-play-17-builder

AI Observability builder — Application Insights distributed tracing, KQL query library for AI metrics, Azure Workbooks dashboards, alerting rules, and FinOps telemetry.

• operational-excellence
• cost-optimization
• reliability

fai-play-17-reviewer

AI Observability reviewer — telemetry coverage audit, KQL query accuracy, dashboard UX review, alert threshold calibration, and PII-in-logs verification.

• operational-excellence
• security

fai-play-17-tuner

AI Observability tuner — Log Analytics commitment tier selection, sampling rate config, alert threshold calibration, dashboard refresh optimization, and retention policy tuning.

• cost-optimization
• operational-excellence

fai-play-18-builder

Prompt Optimization builder — prompt versioning with semantic versions, A/B testing framework, Azure Prompt Flow, template engine with variable injection, and evaluation-driven iteration.

• performance-efficiency
• cost-optimization
• reliability

fai-play-18-reviewer

Prompt Optimization reviewer — version management audit, A/B test methodology review, template injection safety, evaluation pipeline validation, and prompt quality assessment.

• security
• responsible-ai

fai-play-18-tuner

Prompt Optimization tuner — system message clarity, few-shot example selection, A/B test config, template variable defaults, and prompt cost-quality trade-off analysis.

• cost-optimization
• performance-efficiency

fai-play-19-builder

Edge AI builder — Phi-4 on-device inference, ONNX quantization (INT4/INT8), Azure IoT Hub fleet management, offline-first architecture, and cloud fallback patterns.

• reliability
• performance-efficiency
• cost-optimization

fai-play-19-reviewer

Edge AI reviewer — quantized model quality audit, offline resilience testing, fleet rollout verification, cloud fallback review, and device security checks.

• reliability
• security

fai-play-19-tuner

Edge AI tuner — quantization level selection, on-device latency optimization, sync schedule config, cloud fallback thresholds, and per-device cost analysis.

• cost-optimization
• performance-efficiency

fai-play-20-builder

Real-Time Analytics builder — Event Hub partitioned ingestion, Stream Analytics windowing, LLM-powered anomaly explanation, multi-signal scoring, and live dashboards.

• performance-efficiency
• reliability

fai-play-20-reviewer

Real-Time Analytics reviewer — ingestion reliability audit, windowing correctness, anomaly detection accuracy, scoring logic review, and alert quality verification.

• reliability
• security

fai-play-20-tuner

Real-Time Analytics tuner — window size optimization, anomaly threshold calibration, baseline window selection, alert severity rules, and streaming cost analysis.

• cost-optimization
• performance-efficiency

fai-play-21-builder

Agentic RAG builder — autonomous retrieval agent, multi-source fusion (Search+web+DB), iterative query refinement, citation pipeline, and reflection-based quality gates.

• reliability
• performance-efficiency

fai-play-21-reviewer

Agentic RAG reviewer — retrieval autonomy audit, source selection review, iteration limit verification, citation accuracy check, and reflection quality assessment.

• reliability
• responsible-ai