How hardikdhawan2904/mva-azure works
MVA is a multi-agent data pipeline: a user uploads CSV/Excel data to the orchestrator, which deduplicates or versions it, invokes schema quality/classification, then structural profiling and readiness analysis, and optionally analytics. The README documents the major capabilities and service handoffs; sampled pipeline excerpts support several internal stages. Some service-to-service details and unsampled integrations are documented only at the README level, so edges below avoid inferring finer-grained calls.
Main entry points: nodes/pipeline.py Profiling Pipeline · nodes/pipeline.py Pipeline Orchestrator · nodes/pipeline.py Analytics Pipeline · nodes/pipeline.py Schema Service
Components 19
Orchestration
- Pipeline Orchestrator
Agent-Orchestrator/app/agents/orchestration_agent/nodes/pipeline.py - Dataset Registry
Agent-Orchestrator/app/services/dataset_registry/dataset_registry.py
Schema Intelligence
- Schema Service
Schema-Intelligence-Layer/app/agents/schema_intelligence_agent/nodes/pipeline.py - Quality Gate
Schema-Intelligence-Layer/app/services/quality_validator.py - Domain Classification
Schema-Intelligence-Layer/app/agents/classification_agent/nodes/pipeline.py
Data Profiling
- Profiling Pipeline
Data-Profiling-Agent/app/agents/data_profiling_agent/nodes/pipeline.py - Structural Profiling
Data-Profiling-Agent/app/services/profiling/column_profiler.py - Hierarchy Inference
Data-Profiling-Agent/app/services/hierarchy/candidate_builder.py - Quality Rules
Data-Profiling-Agent/app/services/rules/rule_engine.py - AI Readiness
Data-Profiling-Agent/app/services/readiness/readiness_engine.py - Chart Generation
Data-Profiling-Agent/app/services/charts/candidate_generator.py - Rule Suggestions
Data-Profiling-Agent/app/services/rules/suggestion_generator.py
Analytics
- Analytics Pipeline
Analytics-Agent/app/agents/analytics_agent/nodes/pipeline.py - Analytics Capabilities
Analytics-Agent/app/services/capability_resolution/analytics_capability_resolver.py - Analytics Planning
Analytics-Agent/app/services/planning/analytics_planner.py - Analytics Tools
Analytics-Agent/app/services/tools/analytics_tool.py - Conversation Memory
Analytics-Agent/app/services/database.py
Shared Persistence
- Shared Postgres
Other components
- Dataset User
Connections 21
- Dataset User → Pipeline Orchestrator: uploads data
README.md - Pipeline Orchestrator → Dataset Registry: checks identity
Agent-Orchestrator/app/agents/orchestration_agent/nodes/pipeline.py - Pipeline Orchestrator → Schema Service: invokes classification
README.md - Schema Service → Quality Gate: validates quality
Schema-Intelligence-Layer/app/agents/schema_intelligence_agent/nodes/pipeline.py - Schema Service → Domain Classification: classifies dataset
Schema-Intelligence-Layer/app/agents/schema_intelligence_agent/nodes/pipeline.py - Pipeline Orchestrator → Profiling Pipeline: invokes profiling
README.md - Domain Classification → Profiling Pipeline: provides domain and descriptions
README.md - Profiling Pipeline → Structural Profiling: profiles columns
Data-Profiling-Agent/app/agents/data_profiling_agent/nodes/pipeline.py - Profiling Pipeline → Hierarchy Inference: builds hierarchy
README.md - Profiling Pipeline → Quality Rules: evaluates rules
README.md - Profiling Pipeline → AI Readiness: scores readiness
Data-Profiling-Agent/app/services/readiness/readiness_engine.py - Profiling Pipeline → Chart Generation: generates charts
Data-Profiling-Agent/app/agents/data_profiling_agent/nodes/pipeline.py - Profiling Pipeline → Rule Suggestions: proposes rules
README.md - Pipeline Orchestrator → Analytics Pipeline: optionally invokes
README.md - Analytics Pipeline → Analytics Capabilities: resolves capabilities
Analytics-Agent/app/agents/analytics_agent/nodes/pipeline.py - Analytics Pipeline → Analytics Planning: plans analysis
Analytics-Agent/app/agents/analytics_agent/nodes/pipeline.py - Analytics Pipeline → Analytics Tools: uses analysis tools
Analytics-Agent/app/agents/analytics_agent/nodes/pipeline.py - Schema Service → Shared Postgres: persists metadata
README.md - Profiling Pipeline → Shared Postgres: persists profiles
README.md - Analytics Pipeline → Conversation Memory: loads and saves turns
Analytics-Agent/app/services/database.py - Conversation Memory → Shared Postgres: stores turns
Analytics-Agent/app/services/database.py