Merge pull request #85 from anfredette/refactor

refactor: Reorganize data/ directory into benchmarks, configuration, and archive

Author: anfredette

.gitignore

@@ -68,18 +68,13 @@ logs/
 # Data
 *.sql
 !scripts/schema.sql
-data/benchmarks_redhat_performance.json
-data/benchmarks_GuideLLM.json
+data/benchmarks/performance/benchmarks_GuideLLM.json
 
 # Presentation images (large files, regeneratable)
-data/benchmarks/models/presentation/
+data/archive/presentation/
 
 # Old branding (replaced by neuralnav-*)
 docs/ai_assistant_*
 
 # macOS system files (all directories)
 **/.DS_Store
-
-# Redundant/unused data files (identified in data audit)
-data/slo_ranges_from_benchmarks.json
-data/research/benchmark_slo_ranges.json

CLAUDE.md

@@ -48,8 +48,8 @@ SIMULATOR_FULL_IMAGE := $(REGISTRY)/$(REGISTRY_ORG)/$(SIMULATOR_IMAGE):$(SIMULAT
 OLLAMA_MODEL ?= qwen2.5:7b
 KIND_CLUSTER_NAME ?= neuralnav
 
-PGDUMP_INPUT ?= data/integ-oct-29.sql
-PGDUMP_OUTPUT ?= data/benchmarks_GuideLLM.json
+PGDUMP_INPUT ?= data/benchmarks/performance/integ-oct-29.sql
+PGDUMP_OUTPUT ?= data/benchmarks/performance/benchmarks_GuideLLM.json
 
 BACKEND_DIR := backend
 UI_DIR := ui
@@ -419,32 +419,32 @@ db-init: db-start ## Initialize PostgreSQL schema
 
 db-load-synthetic: db-start ## Load synthetic benchmark data (appends)
 	@printf "$(BLUE)Loading synthetic benchmark data...$(NC)\n"
-	@uv run python scripts/load_benchmarks.py data/benchmarks_synthetic.json
+	@uv run python scripts/load_benchmarks.py data/benchmarks/performance/benchmarks_synthetic.json
 	@printf "$(GREEN)✓ Synthetic data loaded$(NC)\n"
 
 db-load-blis: db-start ## Load BLIS benchmark data (appends)
 	@printf "$(BLUE)Loading BLIS benchmark data...$(NC)\n"
-	@uv run python scripts/load_benchmarks.py data/benchmarks_BLIS.json
+	@uv run python scripts/load_benchmarks.py data/benchmarks/performance/benchmarks_BLIS.json
 	@printf "$(GREEN)✓ BLIS data loaded$(NC)\n"
 
 db-load-estimated: db-start ## Load estimated performance benchmarks (appends)
 	@printf "$(BLUE)Loading estimated performance data...$(NC)\n"
-	@uv run python scripts/load_benchmarks.py data/benchmarks_estimated_performance.json
+	@uv run python scripts/load_benchmarks.py data/benchmarks/performance/benchmarks_estimated_performance.json
 	@printf "$(GREEN)✓ Estimated data loaded$(NC)\n"
 
 db-load-interpolated: db-start ## Load interpolated benchmark data (appends)
 	@printf "$(BLUE)Loading interpolated benchmark data...$(NC)\n"
-	@uv run python scripts/load_benchmarks.py data/benchmarks_interpolated_v2.json
+	@uv run python scripts/load_benchmarks.py data/benchmarks/performance/benchmarks_interpolated_v2.json
 	@printf "$(GREEN)✓ Interpolated data loaded$(NC)\n"
 
 db-load-guidellm: db-start ## Load GuideLLM benchmark data (appends)
 	@printf "$(BLUE)Loading GuideLLM benchmark data...$(NC)\n"
-	@if [ ! -f data/benchmarks_GuideLLM.json ]; then \
-		printf "$(RED)✗ data/benchmarks_GuideLLM.json not found$(NC)\n"; \
+	@if [ ! -f data/benchmarks/performance/benchmarks_GuideLLM.json ]; then \
+		printf "$(RED)✗ data/benchmarks/performance/benchmarks_GuideLLM.json not found$(NC)\n"; \
 		printf "$(YELLOW)Run 'make db-convert-pgdump' first to create it from a pg_dump file$(NC)\n"; \
 		exit 1; \
 	fi
-	@uv run python scripts/load_benchmarks.py data/benchmarks_GuideLLM.json
+	@uv run python scripts/load_benchmarks.py data/benchmarks/performance/benchmarks_GuideLLM.json
 	@printf "$(GREEN)✓ GuideLLM data loaded$(NC)\n"
 
 db-convert-pgdump: db-start ## Convert PostgreSQL dump to JSON format

Makefile

@@ -62,7 +62,7 @@ async def list_use_cases():
 async def get_benchmarks():
     """Get all 206 models benchmark data from opensource_all_benchmarks.csv."""
     try:
-        csv_path = _get_data_path() / "benchmarks" / "models" / "opensource_all_benchmarks.csv"
+        csv_path = _get_data_path() / "benchmarks" / "accuracy" / "opensource_all_benchmarks.csv"
 
         if not csv_path.exists():
             logger.error(f"Benchmark CSV not found at: {csv_path}")
@@ -95,7 +95,7 @@ async def get_priority_weights():
     when setting initial weights based on priority dropdowns.
     """
     try:
-        json_path = _get_data_path() / "priority_weights.json"
+        json_path = _get_data_path() / "configuration" / "priority_weights.json"
 
         if not json_path.exists():
             logger.error(f"Priority weights config not found at: {json_path}")
@@ -137,7 +137,7 @@ async def get_weighted_scores(use_case: str):
                 detail=f"Invalid use case: {use_case}. Valid options: {list(use_case_to_file.keys())}",
             )
 
-        csv_path = _get_data_path() / "business_context" / "use_case" / "weighted_scores" / filename
+        csv_path = _get_data_path() / "benchmarks" / "accuracy" / "weighted_scores" / filename
 
         if not csv_path.exists():
             logger.error(f"Weighted scores CSV not found at: {csv_path}")

README.md

@@ -24,7 +24,7 @@ def _calculate_percentile_value(min_val: int, max_val: int, percentile: float =
 
 def _get_slo_workload_path() -> Path:
     """Get path to the SLO workload config file."""
-    return Path(__file__).parent.parent.parent.parent.parent / "data" / "usecase_slo_workload.json"
+    return Path(__file__).parent.parent.parent.parent.parent / "data" / "configuration" / "usecase_slo_workload.json"
 
 
 @router.get("/slo-defaults/{use_case}")

backend/src/api/routes/reference_data.py

@@ -112,7 +112,7 @@ def __init__(self, data_path: Path | None = None):
             data_path: Path to model_catalog.json
         """
         if data_path is None:
-            data_path = Path(__file__).parent.parent.parent.parent / "data" / "model_catalog.json"
+            data_path = Path(__file__).parent.parent.parent.parent / "data" / "configuration" / "model_catalog.json"
 
         self.data_path = data_path
         self._models: dict[str, ModelInfo] = {}

backend/src/api/routes/specification.py

@@ -73,7 +73,7 @@ def __init__(self, data_path: Path | None = None):
             data_path: Path to slo_templates.json
         """
         if data_path is None:
-            data_path = Path(__file__).parent.parent.parent.parent / "data" / "slo_templates.json"
+            data_path = Path(__file__).parent.parent.parent.parent / "data" / "configuration" / "slo_templates.json"
 
         self.data_path = data_path
         self._templates: dict[str, SLOTemplate] = {}

backend/src/knowledge_base/model_catalog.py

@@ -19,7 +19,7 @@
 # Base path for weighted scores CSVs
 # Path: quality/usecase_scorer.py -> recommendation -> src -> backend -> project root -> data
 DATA_DIR = os.path.join(os.path.dirname(__file__), "..", "..", "..", "..", "data")
-WEIGHTED_SCORES_DIR = os.path.join(DATA_DIR, "business_context", "use_case", "weighted_scores")
+WEIGHTED_SCORES_DIR = os.path.join(DATA_DIR, "benchmarks", "accuracy", "weighted_scores")
 
 
 class UseCaseQualityScorer:

backend/src/knowledge_base/slo_templates.py

@@ -90,6 +90,7 @@ def _load_slo_ranges(self) -> dict:
         config_path = (
             Path(__file__).parent.parent.parent.parent
             / "data"
+            / "configuration"
             / "usecase_slo_workload.json"
         )
         try: